[28] magnetom flash_28_jan 2004

FLASH

Issue no. 1.2004ISMRM Edition

Content

MR Spectroscopy with syngoPage 8

Perfusion MRI/MRS forBrain TumorsPage 22

3T MR SpectroscopyPage 48, 50

Case Reports:Prostate MRSPage 58

The Potential of 1H MRS of the BreastPage 64

31P MR Spectroscopy of the HeartPage 66, 74

MR Spectroscopy ofProstate Carcinoma

www.siemens.com/medical

Pric

e 1

0,-

EU

R

MAGNETOM

28

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

2 www.siemens.com/magnetom-world MAGNETOM FLASH 1/2004

Wednesday, June 23

6:00 – 8:00 P.M. Welcome Reception

Thursday, June 24

New paradigms

8:30 – 8:40 A.M. Welcome

8:40 – 9:00 A.M. MAGNETOM World meets again…

9:00 – 9:20 A.M. Why Tim – How did it change the applications?

9:20 – 9:40 A.M. Parallel Imaging with Tim

9:40 –10:40 A.M. Whole Body Imaging – Is it already a clinical reality?

10:40 –11:00 A.M. Coffee Break

11:00 –12:00 P.M. The impact of Tim at NYU

12:00 –12:30 P.M. The technology behind Tim

12:30 – 1:00 P.M. How will Tim affect the future?

1:00 – 2:15 P.M. Lunch

2:15 – 2:30 P.M. Exciting Applications in the MAGNETOM World

New advances

2:30 – 3:00 P.M. Stroke

3:00 – 3:30 P.M. Coffee Break

3:30 – 4:00 P.M. Routine iPAT Neuro Applications

4:00 – 4:30 P.M. Diffusion Tensor Imaging: clinical opportunities

4:30 – 5:30 P.M. Hands-on: Neuro

7:00 P.M. Social Event

Friday, June 25

New applications

8:30 – 9:00 A.M. Ultra High-Field Club News

9:00 – 9:30 A.M. Prostate Spectroscopy

9:30 – 9:50 A.M. 3T Cartilage Imaging and its clinical value

9:50 –10:10 A.M. Breast MR-Perfusion

10:10 –10:30 A.M. Kidney Perfusion


10:45 –12:30 P.M. Hands-on:Kidney PerfusionBreast MR-PerfusionProstate Spectroscopy

12:30 – 1:30 P.M. Lunch

1:30 – 1:45 P.M. Exiting Applications in the MAGNETOM World

1:45 – 2:15 P.M. CMR with the 1st Tim system and the MAGNETOM Trio

2:15 – 2:45 P.M. Cardiomyopathy Assessment with CMR

2:45 – 3:15 P.M. CMR in private practice

3:15 – 4:15 P.M. Hands-on: Cardiac

4:15 – 4:30 P.M. Wrap up

6:00 – 8:00 P.M. Welcome Reception Cardiac MR Ambassador Meeting

EVENTSMAGNETOM WORLD SUMMIT

Rottach-Egern, Tegernsee, Germany

June 23-27, 2004

3rd MAGNETOM World Summit 2004Cardiac MR Ambassador Meeting 2004

3rd MAGNETOM World Summit 2004

Dear MAGNETOM User,

We cordially invite you to join ourcommunity and attend the 3rd MAGNETOM World Summit andCardiac MR Ambassador Meeting.

The community of MAGNETOMUsers has been growing steadilyworldwide over the last years. Byproviding a platform for socializingand exchanging information, wewant to create the path for trand-setting developments in MR.

Both metings will take place inone of the most beautiful areas ofcentral Europe. We look forward tooffering you a traditional Bavarianwelcome at the picturesque Tegern-see, one of Germany’s hiddentreasures.

We look forward to seeing youthere.

With warmest regards,

Dr. Heinrich Kolem Dr. Bernd Montag President MR VP MR Marketing

Registration fee: Euro 500.00

MAGNETOM FLASH 1/2004 3

Friday, June 25

1:45 – 4:15 P.M. Cardiovascular MR at the MAGNETOM World Summit

6:00 – 8:00 P.M. Welcome Reception

Saturday, June 26

8:00 – 8:30 A.M. Registration

8:30 – 9:15 A.M. Siemens TalksSiemens update on products and WIPs

9:15 –10:00 A.M. Customer TalksHow I do a myocardial perfusion exam?


10:45 – 1:00 P.M. Working groups in parallel sessions and live demos

• Sequences and methods: k-space coverage strategies

• Roles of cardiac CT and MR as non invasive diagnostic tools

• Pediatric imaging

• Live demos on syngo 2004A/2004V, Argus and Vessel View

1:00 – 2:30 P.M. Lunch

2:30 – 4:00 P.M. Working groups in parallel sessions and CMR hotline

• CMR with the MAGNETOM Family: system differences

• Myocardial perfusion

• Vessel wall/plaque imaging

• CMR application hotline

4:00 – 4:30 P.M. Coffee Break

4:30 – 6:00 P.M. Customer Talks and CMR Young Investigator Award

7:00 P.M. Social Event

Sunday, June 27

8:30 –10:00 A.M. Customer Talks


10:30 –12:20 P.M. Summary of working groups

12:20 P.M. Closing remarks

12:30 P.M. Adjourn

EVENTSMAGNETOM WORLD SUMMIT

Cardiac MR Ambassador Meeting 2004

Registration fee: Euro 500.00

Hotel Information

This hotel will host the meeting!Dorint Seehotel Überfahrt Überfahrtstraße 10, D-83700 Rottach-EgernPhone: +49 (0)8022 669-0Fax: +49 (0)8022 669-1000

Single Occupancy EUR 175.00Double Occupancy EUR 221.00www.dorint.com/tegernsee

Hotel Bachmair am See Seestraße 47, D-83700 Rottach-EgernPhone: +49 (0)8022 272-0Fax: +49 (0)8022 272-790

Single Occupancy EUR 140.00Double Occupancy EUR 190.00www.bachmair.de

Fax: +49 (0) 91 31 84-21 42

Mailing address:

ATTN: Christina SpillerSiemens AG, Medical SolutionsCorporate CommunicationsHenkestr. 127, D-91052 ErlangenGermany

MAGNETOM World Summit Registration Form

Please complete this form below or visit us at www.siemens.com/magnetom-world for online registration.

If several people from your institution will be participating, please complete this form for each attendee.

Forms must be received before May 28, 2004.

You may either fax or mail this registration form.

Fax: +49 (0)9131 84-2142 Mailing address on the back page.

Title/First Name

Last Name

Institution

Mailing address

(Include country)

Telephone

(Include country code)

E-mail address

I will attend the 3rd MAGNETOM World Summit, June 23 – 25

Cardiac MR Ambassador Meeting, June 25 – 27

Hotel Reservation Yes No

Arrival Date Departure Date

Single Occupancy Double Occupancy (Shared with)

Preferred Hotel* Dorint Seehotel Überfahrt

Hotel Bachmair am See

Please indicate specific issues that you would like to see addressed

by Siemens during this session.

Please verify the visa requirements by contacting your local consulate.

All changes/amendments have to be made to

[email protected].

*Hotel reservation changes are always subject to availability.


CONTENT

EVENTS

2 3rd MAGNETOM World Summit 2004Cardiac MR Ambassador Meeting 2004

PRODUCT NEWS

8 MR Spectroscopy with syngo MR 2004 A/V:Automation with Flexibility SVS, 2D CSI, and 3D CSI, with Long and Short TE

CLINICAL

14 Clinical MR Spectroscopy: A Primer

16 Metabolite Ratios in Clinical MR Spectroscopy

22 Perfusion MRI and MRS for Brain Tumors

32 MR Spectroscopy Precision and Repeatability: Evaluation of Brain CSI Data

36 1H-MRSI Guided Surgery of Brain Tumors

40 Case Report: Gliomatosis Cerebri

42 Spectroscopy in Differential Diagnosis of an Intracranial Mass Lesion

46 Proton Magnetic Resonance Spectroscopy (MRS) in Primary Pediatric Brain Tumors

48 Case Report 3T: Sjögren-Larsson Syndrome

50 Localized Proton Spectroscopy in Hepatic Encephalopathy: Advantage of 3T-High-Field for Discrimination of Glutamine and Glutamate

6 EDITORIAL87 IMPRESSUM


CONTENT

54 Proton MR Spectroscopic Imaging in the Clinical Evaluation of Prostate Cancer

56 Results from IMAPS Study

58 Case Report: Prostate Carcinoma Stage T3b

60 Prostate Carcinoma Detected with Single Voxel Spectroscopy

62 1H-MR Spectroscopic Imaging of the Human Prostate: from 1.5 to 3T

64 The Potential of 1H MRS of the Breast

66 31P-MR Spectroscopy of the Heart – Current Status and Future Potential

74 31P-Chemical Shift Imaging for Myocardial Infarction

76 Functional investigation of exercising muscle: a non-invasive Magnetic Resonance Spectroscopy – Magnetic Resonance Imaging approach

84 FAQs

LIFE

86 Life

The information presented in MAGNETOM Flash is for illustration only and is notintended to be relied upon by the reader for instruction as to the practice of medicine.Any health care practitioner reading this information is reminded that they must usetheir own learning, training and expertise in dealing with their individual patients.This material does not substitute for that duty and is not intended by Siemens MedicalSolutions to be used for any purpose in that regard.

The drugs and doses mentioned in MAGNETOM Flash are consistent with the approvallabeling for uses and/or indications of the drug. The treating physician bears the soleresponsibility for the diagnosis and treatment of patients, including drugs and dosesprescribed in connection with such use. The Operating Instructions must always bestrictly followed when operating the MR System. The source for the technical data isthe corresponding data sheets.


EDITORIAL

“Life is complex. It has real and imaginary components”

Based on the experience of decades,we have developed new MRSproducts and WIP packages and wehave optimized workflow for clinicalpractice. The contributions of thisissue show how our 1H MRS packagescan add value to clinical MR exams,and how multinuclear capabilites can be used to answer questions ofbasic or clinical research.

New technical developmentssuch as CSI acquisition weighting,Matrix Spectroscopy and displayfunctionality such as transparentmetabolite images translate imme-diately into user benefits. This tech-nical development is on-going:Supporting new MR systems andespecially higher field-strengthsystems, new sequences for makingthe most out of precious measure-ment time, and new post-processingconcepts designed for extracting the relevant information are on ourlist. And since at the end of the daythe user determines the value of ourproduct, supporting you continues to remain the essential last step tosuccess.

Our research, development and customer support activities thusserve a common goal: making yourlife with respect to MRS less complex.

Enjoy reading!

Stefan Roell

On behalf of the MR application developmentMRS team and many product managers andapplication specialists supporting MRS

Happily ignoring any deeper meaning for the moment, spectros-copists like myself may regard thementioned complexity as only thetop of the iceberg of MRS complexity. Nevertheless, despite of all its complexity, life goes on: Researcherscontinue to discover the rich world of MRS techniques and applications,medical doctors find it worthwhile to cope with the extra bit of com-plexity of clinical MRS to establishnew applications. And SiemensMedical Solutions continues to investin it. Why is that so?

Spectroscopy Team


EDITORIAL

G. Heder, Ph.D.MR Application DevelopmentSpectroscopy Team

N. Salibi, Ph.D.R&D Collaborations, USA

J. Ruff, Ph.D.MR Application DevelopmentSpectroscopy Team

S. Roell, Ph.D.MR Application DevelopmentSpectroscopy Team

U. Boettcher, Ph.D.MR Application DevelopmentSpectroscopy Team

E. Weiland, Ph.D.MR Application DevelopmentSpectroscopy Team

M. Vorbuchner, R.T.MR Application DevelopmentSpectroscopy Team


PRODUCT NEWSMR SPECTROSCOPY

MR Spectroscopy with syngo MR 2004 A/V: Automation with Flexibility SVS, 2D CSI, and 3D CSI, with Long andShort TE

workflow, so that no coil changes arerequired for MRS exams and recon-struction times are hardly noticeable.Hence, we minimize not only acquisi-tion times, but also the entire examtime.

Matrix Spectroscopy supports all types of sequences: CSI, SVS andFID acquisitions. CSI datasets of a size of e.g. 32x32x16x1024 complexsamples can be simultaneouslyreceived from four channels. When,for example, 16 receive channels areused, there are still no restrictions on other acquisition parameters suchas TR or spectral resolution. Further-more, a prime development goal hasbeen the robustness of the method:phase coherent signal combinationensures maximum SNR. Due to theself-weighting property of the algo-rithm, the combined data is largelyindependent of the choice of coilelements made by the user. Robust-ness and accuracy has been provenby extensive in vivo testing. Finally,homogenization of CSI data is obtainedeither by normalizing the signals (i)to the very homogeneous sensitivityprofile of the body-coil using pre-scandata, or (ii) to the least-squares normwell known for image normalization.

Lipid Contamination and OuterVolume SaturationLipid signals in the range of 0 ppm to2 ppm are more prominent on shortTE spectra. They may appear in brainspectra from inside the volume ofinterest (VOI) as a result of pathology,

Nouha Salibi1, Ph.D.Stean Roell2, Ph.D.

1 Siemens Medical Solutions,R&D Collaborations, Malvern, PA, USA2 Siemens Medical Solutions,MREA, Erlangen, Germany

on 3 reference images (Fig. 3), freeangulation capability, automatedmeasurement adjustment software,and automated post-processingsoftware. Both SVS and CSI sequenceshave volume-selective schemes thatare based on either the SE (Spin Echo)technique or the STEAM (STimulatedEcho Acquisition Mode) technique.All sequences have built-in variableTEs (echo times), with a minimum TEvalue of 30 ms for the SE-basedsequences and 20 ms for STEAM. InMR spectroscopy, variable TE valuesprovide the ability to control the T2 contrast of spectral peaks in thesame way tissue T2 contrast iscontrolled in MR imaging. Short TEmeasurements are important fordetection of metabolite signals thathave a short T2 decay and are notvisible on long TE spectra [1].

Matrix Spectroscopy

Matrix Spectroscopy supports MRSexams using Tim matrix coils or anyother multiple element receive coil.This unique feature is available withsyngo MR 2004V for the first time. By using matrix coils, greater SNR isgained than with single channel coils– SNR is something we can neverhave enough of in MRS. This SNRgain can be invested in, for example,reducing acquisition time or inincreasing spatial resolution. Further-more, due to its complete integrationwithin Tim technology, Matrix Spec-troscopy ensures that spectroscopyexams are part of the optimized Tim

Over the last few years, MR Spectros-copy (MRS) has become a routinecomponent of clinical MRI examina-tions at many imaging centersaround the world. The latest releaseof Siemens’ syngo MR software isdesigned to offer unprecedentedversatility and ease of use, allowingyou to maximize the clinical useful-ness of MRS at your institution.

MRS Techniques with syngoMR 2004 A/V: More Versatilitythan Ever BeforeProton MR spectroscopy* with syngoMR 2004 A/V includes single-voxelspectroscopy (SVS), as well as multi-voxel 2D and 3D chemical shiftimaging (2D CSI and 3D CSI), withVOI (volume of interest) positioning

* The multinuclear option is available forMAGNETOM Symphony and Sonata systems(including upgrades) with syngo MR 2002B andabove, for MAGNETOM Trio and Allegra with2004A. For the MAGNETOM Avanto system itwill be available with the upcoming SW release.

WIP: The information about this product ispreliminary. The product is under developmentand is not commercially available in the US andits future availability cannot be ensured.

Figure 1 Outer volume saturation (OVS)slabs placed on a transverse brain imageto suppress lipid signal from the scalp.

Figure 2a Positioning of outer volume saturation (OVS) slabs around a 3D CSI VOI of a proton MRS examination in prostate cancer (left). OVS, along withspectral lipid suppression, eliminate lipid contamination inside the VOI. This is clearly demonstrated in the spectral map (right) and on the spectrashown in Fig. 2b.

Figure 2b The image on the left shows a spectrum from a voxel in the lesion(indicated by the short red arrow). It has a high choline peak at 3.2 ppm (yellow arrow) and a reduced citrate signal at 2.6 ppm (green arrow) compared to the spectrum from a voxel in normal tissue (indicated by the long red arrow inthe image on the right). The result shows an excellent lipid suppression.





Figure 3 Positioning of a 2D CSI slice including presumably healthybrain tissue, as well as most of a lesion diagnosed as infiltratinganaplastic astrocytoma; the patient had undergone a resection of thisneoplasm 12 months prior to this MRS follow-up examination.

Figure 4 Spectra from the 2D CSI slice in Fig. 3. A 4 min 39 sec 2DCSI measurement was performed on the same slice at three differentTEs. The last three columns show spectra from the same voxel(shown in the first column of each row) at the TEs indicated along thetop of the figure. The first row represents control spectra. The last two rows are from the lesion. Spectra from the lesion have a reducedNAA/Cr ratio and an elevated Cho/Cr ratio. Short TE spectra from thelesion (second column, last two rows) have elevated myo-inositol(mI) (arrow) compared to the control spectrum (second column, firstrow).

or from outside the VOI as a result ofcontamination when a VOI is positio-ned very close to lipid-containingstructures. Such contamination isalso seen in prostate spectra, and canbe eliminated by carefully placingouter volume saturation (OVS) slabsoutside the volume of interest (Figs.1 and 2a). In addition to OVS, prostateMRS sequences include spectral lipidsuppression. A total of 12 OVS slabscan be positioned. 8 can be placedinteractively by the user and 4 invisibleslabs using the “fully excited VOI”feature.

MRS Data Presentation: Easy to UseThe automated syngo MR 2004 A/Vpost-processing software offers a number of new features for easypresentation of MRS data. Metaboliteratios for a single-voxel spectrum areautomatically calculated and presentedin a “table of results” format. CSI data can be displayed as singlespectra, spectral maps, metaboliteimages, or images of metaboliteratios (Figs. 4, 5, and 6). Integralvalues of metabolite peaks (or theirratios) for individual voxels can beoverlaid as a matrix onto an appro-priate MR image or onto a “metaboliteimage”.

2D CSI with syngo MR 2002 A/V

Voxel position TE = 30 ms TE = 144 ms TE = 288 ms



Figure 5 Top row: spectral map(left) and matrix of NAA integralvalues (right) generated from the 2D CSI data set (TE = 144 ms) demonstrate distribution of metabo-lites over the region of interest.Bottom row: map of the Cho/NAAratio overlaid onto the correspondingmetabolite image (right), and theNAA peak integral map overlaid ontothe corresponding transverse image(left). It is clearly shown that NAA is considerably reduced within thelesion and in the immediately surrounding areas. The MRS results,along with other findings, suggestpresence of residual tumor.



Figure 6 This figure illustrates some features of the syngo MR 2002A/V3D CSI technique, including coverage, resolution, short TE acquisition,reduced measurement time, and a variety of ways to display the corre-sponding MRS data. This 3D CSI data set was acquired on the brain of a healthy volunteer with TR = 1500 ms, TE = 30 ms, and a measurementtime of 7 min 46 sec, using weighted k-space sampling. The volume ofinterest (60 mm x 80 mm x 60 mm) is angled by 40 degrees from thetransverse to the coronal orientation. Of the eight 3D partitions measu-red, four are within the volume of interest as demonstrated above. Each partition has a thickness of 15 mm and an in-plane resolution of 10 mm x 10 mm. The first row indicates the position of each of the fourpartitions that corresponds to the data shown in the same column. The second row shows spectral maps corresponding to each partition.The third row illustrates the display of the NAA integral values over thecorresponding NAA metabolite images. The fourth row displays highquality short TE spectra from voxels in the third row (indicated by arrows).

Oblique 3D CSI with syngo MR 2004 A/V

TE = 30 ms. Measurement time: 7 min 46 sec

Conclusion

MR Spectroscopy with the latestrelease of the syngo MR softwareoffers unprecedented clinical utilitythrough automation and flexibility.The full range of MRS techniques(SVS, 2D CSI, and 3D CSI) can beperformed, with both long and shortTEs. The MRS data can be displayedin several user-friendly formats,making it easier than ever to integrateMRS into your clinical MR imagingroutine.

Acknowledgements:

Special thanks to Dr. A. Heerschap, Dr. T. Scheenen, Dr. D. Klomp et al., University Hospital Nijmegen, St. Radboud, NL,for the MRS data of Figure 2, and to Dr. E. Knopp and Dr. M. Law, NYU Medical Center, for the MRS data of Figures 3, 4, and 5.

It’s a whole new way to think about MR.

Going beyond incremental enhancements in

magnet technology and gradient power to

something radically different. A transforming

technology that shifts the spotlight onto a

new and tremendously innovative RF system

and matrix coil design. Tim technology

is a unique combination of features that

will launch the next era in powerful MR

advances. From revolutionary acquisition

quality to breakthrough applications and

remarkable profitability. It’s opening up

new worlds of possibility.

• Tim is a revolutionary. Select exams,

not coils. [76 x 32] Up to 76 seamlessly

integrated matrix coil elements and up

to 32 RF channels combined to create one

Total imaging matrix.

• Tim sees all. Local becomes total.

True 205 cm (6′ 9″) seamless, whole-body

anatomical coverage with no coil

reconfiguration or patient repositioning.

• Tim knows no boundaries. Parallel in all

directions. Head to toe, front to back, and

side to side for unlimited Parallel Imaging.

Timwill change MR forever.

case. In certain instances, SVS is astraightforward approach to collectingthe desired MRS information (especi-ally with metabolic brain disorders);it produces a single spectrum from a single volume of 2 cm3 to 8 cm3

in size. In other instances, however,SVS may not be optimal for investi-gating a lesion that is too large to befully contained within the voxel, or a lesion that is too small to affectspectral peaks from a 2 cm3 voxel. 2Dor 3D CSI multi-voxel techniquesallow better coverage of one largelesion or multiple lesions, and allowhigher spatial resolution than SVS,which is needed for the investigationof regional variations within the VOI.Moreover, CSI measurement timesare currently comparable to SVSmeasurement times. In the examplesshown in Figures 3-4, the CSI appro-ach allows examination of the lesion,as well as of the tissue surroundingthe lesion, from the same CSI dataset.

MRS at Various TEs

In MR spectroscopy, variable TEvalues provide the ability to controlthe “T2 contrast” of spectral peaks inthe same way tissue T2 contrast iscontrolled in MR imaging. Short T2metabolite signals decay faster, andthe corresponding spectral peaks are not seen on long TE spectra. As illustrated in Figure 5, detection ofsuch metabolites requires short TEmeasurements. The major healthybrain metabolite peaks that are seenon long TE spectra include N-acetylaspartate (NAA) at 2.02 ppm and 2.6 ppm, total choline (Cho) at 3.20ppm, and total creatine (Cr) at 3.02ppm and 3.9 ppm. Short TE spectracontain additional peaks, whichinclude glutamine and glutamate(Glx) between 2.05-2.5 ppm and3.65-3.8 ppm, scyllo-inositol (sI) at3.36 ppm, glucose at 3.43 ppm and


CLINICALMRS BRAIN

Nouha Salibi, Ph.D.

Siemens Medical Solutions USA,R&D Collaborations,Malvern, PA, USA

Clinical MR Spectroscopy: A Primer

MRS Techniques

Clinical proton MRS techniques includesingle-voxel spectroscopy (SVS) andmulti-voxel 2D and 3D chemical shiftimaging (2D CSI and 3D CSI). SVS and some CSI sequences use volume-selective schemes that are based oneither the SE (Spin Echo) techniqueor the STEAM (Stimulated EchoAcquisition Mode) technique. BothSE and STEAM have three selectiveradiofrequency (RF) pulses to excitethree orthogonal planes. A spectrumis collected from the volume definedby the intersection of the threeexcited planes. The SE sequence hasone 90° pulse followed by two 180°pulses, whereas STEAM has three 90°pulses. While both sequences yieldthe same metabolic information, SE has twice as much signal. Singlevoxel spectroscopy produces a singlespectrum from a single voxel (Fig. 2)that is typically 8 cm3 in volume,whereas CSI measures spectra frommultiple voxels that are typically 1 cm3-1.5 cm3 in volume. CSI datamay be presented in a variety ofdisplays including individual spectra,spectral maps, or colored metaboliteimages overlaid on anatomicalimages (Figs. 3, 4). The two techni-ques are illustrated in Figures 2-4,where the same lesion has beenexamined with SVS (Fig. 2) and CSI(Figs. 3, 4). The two measurementsyield comparable metabolic differen-ces between spectra from the lesionand from the surrounding tissue.However, the relative changesamong peaks are slightly differentdue to the difference in the relativeamount of healthy tissue containedin the SVS (8 cm3) and the CSI (1.5 cm3) voxels.

Which Technique When?

The choice of MRS technique shouldbe tailored to each particular clinical

MR Spectroscopy as a Clinical Tool

Over the last few years, MR spectros-copy (MRS) has migrated beyondresearch laboratories to become anintegral part of the clinical diagnosticroutine. Although its main currentapplication consists of proton MRS inthe brain, there is a growing interestin using MRS for understanding patho-logy in the prostate, liver, muscle,breast, and heart. Multi-nuclearspectroscopy* promises to extendthe clinical applications of MRS evenfurther. MR images are based on MRsignals from water and fat, which arethe two most abundant metabolitesin the body and generate strong MRsignals that dominate those fromother metabolites. Apart from waterand fat, MRS can detect metabolitesthat are indicative of diseases, andcan help in evaluating the effective-ness of certain therapeutic approa-ches. In many instances, MRS is anattractive, noninvasive approach forresolving ambiguous findings seenon MR images and for monitoring therecovery of tissue under varioustherapeutic approaches.

* The multinuclear option is available forMAGNETOM Symphony and Sonata systems(including upgrades) with syngo MR 2002B andabove, for MAGNETOM Trio and Allegra with2004A. For the MAGNETOM Avanto system itwill be available with the upcoming SW release.

WIP: The information about this product ispreliminary. The product is under developmentand is not commercially available in the US andits future availability cannot be ensured.


CLINICALMRS BRAIN

Figure 1 Spectra from normal braintissue. Small metabolite peaks arenot visible in the presence of a largewater peak (left spectrum). A water-suppressed spectrum on the rightclearly displays the major brain meta-bolite peaks at TE = 144 ms, namelyN-acetyl aspartate (NAA), choline(Cho), and creatine/phosphocreatine(Cr/PCr).

Figure 2 (a), a single voxel is posi-tioned to enclose a low-grade brainstem glioma. The voxel is 8 cm3. The corresponding spectrum (b) isacquired with the SE sequence (TR = 1500 ms and TE = 30 ms). Thespectrum shows reduced NAA (greenarrow) with elevated choline (yellowarrow) and elevated myo-inositol(red arrow) and Glx (orange arrow).

Figure 3 Positioning of a 2D CSI slice including the low-grade brain stemglioma, as well as healthy brain tissue. Some CSI results are shown in Fig. 4.

Figure 4 2D CSI data set acquired with TR = 1500 ms, TE = 30 ms, and a spatial resolution of 1.5 cm3. CSI results are shown as a spectral map (b), asspectra from the voxels indicated by red arrows (a), and as a choline meta-bolite image (c) generated from integral peak values. Elevated choline withinthe lesion is seen in the spectral map (b) and in the metabolite image (c) (red area). The upper spectrum in (a) is from the lesion and shows reduced NAA (green arrow), and elevated Cho (yellow arrow), mI (red arrow), and Glx,as compared to the lower spectrum in (a) which is from surrounding tissue.

3.8 ppm, and myo-inositol (mI) at 3.56 ppm and 4.06 ppm. Figure 5also illustrates the dependence ofpeak metabolite ratios on TE. At TE =144 ms, the choline peak is higherthan creatine, whereas at TE = 30 ms,choline is lower than creatine due to the fact that creatine has a shorterT2 and decays faster than choline.

MRS and Magnetic FieldStrengthMR spectroscopy is currently performedon clinical and research scanners atmagnetic field strengths of 1 Tesla,1.5 Tesla, and 3 Tesla. Spectroscopyat fields higher than 1.5T has beendriven by the demand for improvedsensitivity and better spectral resolu-tion (i.e. chemical shift dispersion),which in turn allow for more reliablequantitation of MRS spectra. However,clinical MRS at higher fields presentsnew challenges, including increasedchemical shift misregistration, andincreased magnetic susceptibilityeffects that reduce resolution andsensitivity. New technical develop-ments to overcome such challengesare currently being investigated andimplemented in order to optimize the advantages of higher field MRS.

Conclusion

In recent years, considerable techno-logical advances in MR scanner hard-ware and software have allowed thedevelopment of new and sophistica-ted approaches to MR spectroscopy.The currently available SVS and CSItechniques offer flexibility, automa-tion, and a variety of features thatcan be tailored to specific clinical MRexaminations in order to make MRspectroscopy an integral part of yourclinical MRI routine.

Figure 5 Spectra from the same 8 cm3 single voxel in healthy brain tissue, acquired with the SEsequence at TE = 144 ms (left) and TE = 30 ms (right). Additionalmetabolite peaks are seen on the short TE spectrum. (Not allpeaks are labeled for clarity).

TE = 144 msNAA

ChoCho

Cr Cr

NAA

NAA

Cr/PCr

Cr/PCr

WaterSignal

Metabolitesignals

Cho

TE = 30 ms

Cr Crml

Glx

Figure 1 Spectra from an 8 cm3 single voxel measurements invarious regions of the brain at 1.5T: Parietal white matter (a),midline gray matter (b), cerebellum (c) and pons (d). All spectrawere acquired with the SE sequence at TE = 30 ms, TR = 1500 msand 128 averages. Visual inspection of these spectra clearlydemonstrates variations in the relative peak amplitudes. Similar variations are reflected in the calculated peak integralvalues and metabolite ratios.

2-Regional Variations of Metabolite Ratios

Figures 1 and 2 demonstrate varia-tions in spectral patterns with voxelposition and with TE. Data wereobtained at 1.5T from the brain of ahealthy volunteer with TR = 1500 ms,TE = 30 ms (Fig. 1) and TE = 144 ms(Fig. 2) in white matter, gray matter,cerebellum and pons. The observedvariations imply that metaboliteratios from a voxel containing patho-logy should be compared to meta-bolite ratios measured with the same parameters at a similar voxelposition in healthy volunteers or in an unaffected contra-lateral area of the brain, if possible.

16 www.SiemensMedical.com/MAGNETOM-World MAGNETOM FLASH 1/2004

CLINICALMRS BRAIN

Nouha Salibi, Ph.D.

Siemens Medical Solutions USA,R&D Collaborations,Malvern, PA, USA

Metabolite Ratios in Clinical MRSpectroscopy

require additional correction. Although this method is practical touse clinically, its success depends onthe consistent use of the same measure-ment technique, same parametersand identical voxel position on allsubjects. As demonstrated below,metabolite ratios vary with measure-ment parameters such as TR (Fig. 1)and TE (Fig. 5, table 1) and withregions of the brain (Figs. 1 and 2).Additionally, ratios are typically takenwith respect to creatine (Cr), which is the most stable metabolite in the brain. However certain diseasesdo affect creatine levels in the brain, as demonstrated in the case studypresented below.

1-Quantification in ClinicalMR Spectroscopy

MR spectroscopy (MRS) is an establi-shed and accurate analytical tool inchemistry and biochemistry where itis used under optimized and controlledexperimental conditions. The abilityof MRS to provide reliable absolutequantification has not materializedwith clinical MR Spectroscopy becauseit is not currently practical to incorpo-rate absolute quantification intoroutine clinical practice. Time andeffort are necessary to meet technicaland clinical challenges, which includerelatively noisy spectra and poorlyseparated peaks at 1T and 1.5T, the use of an internal or externalreference, and the need for correc-ting many parameters that affectspectral peak intensities such as coilloading, partial volume effects,regional susceptibility effects andrelaxation times. Of these challenges,only those related to S/N and peakseparation may benefit from MRS athigher field strength.

Many approaches to clinicalinterpretation of spectra have beenproposed as an alternative to absolu-te quantification. In certain instan-ces, visual inspection of spectra orqualitative interpretation may providethe clinical answer. A more commonpractice is the comparison of meta-bolite ratios from pathological tissueto those from identical normal braintissue of healthy volunteers. Meta-bolite ratios are readily provided bythe scanner software and do not

a

b

c

d


CLINICALMRS BRAIN

Figure 3 An 8 cm3 voxel is positionedmostly in white matter to measure thespectra in figure 4.

Figure 4 Spectra from the voxel shown in figure 3, acquired with the SE sequence at TR = 1500 ms, and 3000 ms (TE = 144 ms)are displayed on the same scale. The longerTR spectrum has higher metabolite peaksdue to the longer time available for T1recovery of the magnetization.

3-Metabolite Ratios at Various TR

For optimum S/N per unit time, clinicalMR spectra are typically measuredwith a TR of 1500 ms to 3000 ms forboth single voxel spectroscopy (SVS)and chemical shift imaging (CSI).Figure 4 shows spectra obtained atthese TR values with a TE of 144 msfrom the voxel defined in Figure 3.The metabolite ratios derived frompeak integral values of these spectraare NAA/Cr = 2.21 and 2.0, and

Cho/Cr = 1.20 and 1.08 at TR = 1500ms and 3000 ms respectively. Thelonger TR spectrum has highermetabolite peaks due to the longertime available for T1 recovery of themagnetization. The decrease inmetabolite ratios with TR is due tothe fact that T1 of creatine (Cr) islonger than T1 of Choline (Cho) andNAA (N-acetyl aspartate). Hence, foraccurate interpretation and compa-rison of spectra, it is best to performroutine MRS exams with the samerepetition time.

Figure 2 Spectra from 8 cm3 single voxel measurements invarious regions of the brain at 1.5T: Parietal white matter (a),midline grey matter (b), cerebellum (c) and pons (d). All spectrawere acquired with the SE sequence at TE = 144 ms, TR = 1500 msand 128 averages. Visual inspection of these spectra clearlydemonstrates variations in the relative peak amplitudes. Similar variations are reflected in the calculated peak integralvalues and metabolite ratios.

a

b

c

d

4-Metabolite Ratios at Various TE’s

N. Salibi1, Ph.D.; Glenn Foster 2, RT,RMR and Richard Nagel2, RT, RMR1 Siemens Medical Solutions USA, Inc,Malvern, PA, USA2 Washington University School ofMedicine, St. Louis, MO, USA

Although clinical brain spectra areroutinely measured at the same TRvalue, it is now common practice tomeasure the same SVS or CSI volume

TR = 1500 ms TR = 3000 ms

NAA

NAA

Cho Cr

CrCho

Figure 7 This figure shows the signal decay of NAA,choline (cho) and creatine (Cr). T2 values were derivedfrom a semi logarithmic plot of peak integral values of the spectra in figure 6 as a function of TE.

Table 1 metaboliteratios at differentvalues of TE from thevoxel in figure 3.


CLINICALMRS BRAIN

at more than one TE value. MultipleTE measurements yield added meta-bolic information and increasedconfidence in the interpretation ofspectra. The experiment described inthis section illustrates the effect of TE on metabolite ratios.

MR spectroscopy measurementswere performed on a 1.5TMAGNETOM Sonata and on the brainof a healthy adult volunteer. A 20 mm x 20 mm x 20 mm voxelwas placed in parietal white matter(Fig. 5); the spin echo (SE) sequencewas used with TE = 30 ms, 60 ms,100 ms, 144 ms, 200 ms, 250 msand 288 ms. TR (2000 ms) and allother parameters including the shimwere kept constant. Spectra in Figure6 were evaluated with the Siemenspost-processing software.

Peak integral values from Figure 6were used to calculate the metaboliteratios listed in Table1. The ratiosindicate increasing values of NAA/Crand Cho/Cr at longer echo times. Theincrease is due to the fact that creatinehas the shortest T2 relaxation timeand decays faster than NAA andcholine (Fig. 7). This difference in theT2 decay of metabolites also explainsthe fact that shorter TE (30 ms)spectra* contain more metabolitepeaks (i.e. short T2 metabolites) thanlonger TE (144 ms and 288 ms)spectra (Fig. 6). In spectroscopy, as in imaging, TE may be used to mani-pulate relative signal intensities orcontrast. Hence interpretation ofclinical spectra and metabolite ratios,like the interpretation of MR images,has to take into account the measure-ment echo time.

Figure 5 Position of the voxel used tomeasure metabolite ratios at various TEs.

Figure 6 Spectra from parietal white matter of adult brain obtained at 1.5Twith the SE sequence at various TEs. All other parameters including the shimwere the same. The effect of TE on metabolite ratios is easily seen by comparingrelative amplitudes of choline and creatine peaks at various TEs.

TE(ms) NAA/Cr Cho/Cr

30 1.70 0.84

144 2.13 1.14

288 3.49 1.60

TE = 30 ms TE = 60 ms TE = 100 ms

TE = 200 ms TE = 250 ms TE = 288 ms

TE = 144 msNAA

ChoCr

NAA

ChoCr

TR = 1500 ms

NAAT2 = 135 ms

Cho Cr

ChoT2 = 377 ms

CrT2 = 205 ms

1.0

0.8

0.6

0.4

0.2

0.00 500 1000 1500 2000


CLINICALMRS BRAIN

5-Clinical Interpretation of Metabolite Ratios: A Case Study

N. Salibi1, Ph.D., E.A. Knopp2, M.D.,Meng Law2, M.D., 1 Siemens Medical Solutions USA, Inc,Malvern, PA, USA2 New York University school ofMedicine, New York, NY, USA

A 3D CSI data set was obtained on a 45 year-old woman with a 10-dayhistory of right arm and leg weak-ness, diagnosed with a stroke affec-ting the left middle cerebral arteryvascular territory. Spectra wereobtained at 1.5 T from a 3D volumemeasuring 80 mm x 70 mm x 40mm. The CSI spin echo sequence wasused with a TE of 144 ms, TR of 1500ms, a spatial resolution of 10 mm x10 mm x 10 mm and a measurementtime of 8 min 20 sec. MRS data wereprocessed with the Siemens post-processing software.

Figure 8 shows the selected 3DVOI (volume of interest) positionedover the affected area of the brain as seen on the MR images. Data fromthe slice indicated in the graph arerepresentative of the complete 3Ddata set. Transparent** color imagesof Cho/Cr (Fig. 9) and NAA/Cr (Fig. 10)along with maps of numerical valuesof metabolite ratios show elevatedCho/Cr and NAA/Cr ratios on theaffected left side of the brain compa-red to the normal contra-lateral side.

* For description of MRS techniques and theirapplications in the brain, the reader is referredto page 14 of this issue: Clinical MR Spectroscopy: A Primer.

** Transparent metabolite images are availablewith syngo MR 2004.

Figure 8 Positioning of a 3D CSI slab with a display of the slice used to generate the metabolite images inthe following figures.

Figure 9 A Color metabolite image** of Cho/Cr (left), and a map ofthe numerical values of Cho/Cr (right). Both indicate higher Cho/Cr onthe affected left side of the brain compared to healthy tissue on theright side of the brain.

Figure 10 A color metabolite image** of NAA/Cr (left), and a map of the numerical values of NAA/Cr (right). Both indicate elevatedNAA/Cr on the affected left side of the brain compared to healthytissue on the right side of the brain.


CLINICALMRS BRAIN

This however does not imply anelevation in the NAA and Cho levels.Indeed NAA and choline metaboliteimages (Fig. 11) indicate a reductionof NAA and choline in that area.Furthermore inspection of the spectralmap (Fig. 12) and of individual spectra(Fig. 13) confirms the fact that allmetabolites are reduced on the leftside of the brain; and the higherCho/Cr and NAA/Cr ratios are a resultof creatine being reduced the most in the setting of ischemia. Figures 13and 14 demonstrate the importanceof displaying spectra on the samescale for accurate comparison andinterpretation.

Figure 11 Metabolite images** of NAA (left) and choline (right)indicate reduced metabolite levels on the affected left side of thebrain compared to healthy tissue on the right side of the brain.

** Transparent metabolite images are availablewith syngo MR 2004.

Figure 12 Spectralmap shows overallreduced level of all metabolites onthe affected left sidecompared to the rightside of the brain.


CLINICALMRS BRAIN

Figure 13 Spectra from indicated voxels within the affected side of the brain (left) and from contra-lateral normal voxels (right) are displayed on the same scale. Spectra from the affected side show overallreduction in metabolites and the presence of lipids, an indication of membrane breakdown. Reduction of creatine and NAA is more pronounced than that of choline. The absence of lactate suggests that this is probably a subacute stroke.

Figure 14 The same spectra as in figure 13 are displayed on different vertical scales defined by thehighest peak in each spectrum. Without taking into consideration differences in vertical scales thisdisplay suggests that the spectra on the left show an elevation in choline levels with reduced NAA andcreatine levels, and presence of lipids. The spectra may then be misinterpreted as tumor spectra.

Cr

Cholipids

NAA

CrCho

NAA

Cr

Cho

NAA

Cr

Cho

0.05

0 0

0.15

0.05

00

0.15

lipids

NAA

A 16 x 16 phase-encoding matrixgives an 8 x 8 array of spectra in theVOI with an in plane resolution of 10mm x 10 mm. Although the acquired2D or 3D CSI data set can be measuredwith a higher spatial resolution, tosave time interpolation is sometimesused to reduce the effective voxelsize. Spatial resolution is set byadjusting the FOV and matrix sizewhile keeping the measurement timeat 8 minutes or less. A second 3D (or 2D) CSI sequence is also obtained(time permitting) with TR/TE = 1500ms/30 ms (Table 1).

Clinical Applications ofPerfusion MR and MR Spectroscopy in Brain Tumors

Assessment of Glioma Grade Combined Perfusion MRI and MRSpectroscopy in the clinical settingprovide complementary informationthat can be used for prospectivegrading of gliomas. This has impor-tant clinical utility to the neurosurgeonand neuro-oncologist (Figs. 2-5). Forexample, high-grade gliomas gene-rally have higher relative cerebralblood volume rCBV* measurementsand Cho levels than low gradegliomas [1].

Gliomatosis CerebriGliomatosis cerebri is characterizedby involvement of at least two lobesof the brain by a glial cell tumor ofneuroepithelial origin with relativepreservation of neuronal architecture[2]. Gliomatosis cerebri, which refersto the contiguous involvement ofdifferent regions of the brain, mustbe differentiated from multi-centricglioma, which is defined as multiplefoci of tumor in different sites. Histo-pathologically, there is a lack ofvascular hyperplasia in gliomatosiscerebri; this accounts for the relativelylow rCBV* measurements, mean


CLINICALMRS BRAIN

Meng Law1, M.D.; N. Salibi2, Ph.D.;E.A. Knopp1, M.D.

1 New York University school ofMedicine, New York, NY, USA

2 Siemens Medical Solutions USA,Inc, Malvern, PA, USA

Perfusion MRI and MRS for Brain Tumors

power injector at a flow rate of 3-5ml/sec through the intravenouscatheter (18-22 gauge), the contrastagent injection is immediately followed by a bolus injection of saline(total of 20 ml at the same rate).

Proton MRSI protocol

The volume of interest is confirmedwith scout HASTE (Half-FourierAcquisition Single-Shot Turbo-Spin-Echo) images (TR/TE/excitations = 15 ms/6 ms/1; inversion time of 500ms). Ten 5 mm sections are obtainedwith a 1 minute 15 seconds scan in the axial, coronal, and sagittalplanes. Depending on lesion size, a volume selective 2D or 3D CSIsequence with TR/TE = 1,500 ms/ 144 ms (or 1500ms /30ms) is usedfor MRSI. A measurement at TE = 288 ms was initially included in theprotocol and dropped later. Asdemonstrated in the examples below,this measurement provided no addi-tional clinical information, althoughit contributed to a better understan-ding of the behavior of metabolitesignals at various TEs. The hybridmultivoxel CSI technique uses aPRESS double spin echo scheme forpre-selection of a volume of interest(VOI) that is usually defined to includethe abnormality as well as normal-appearing brain tissue when possible.To prevent strong contribution to the spectra from subcutaneous fatsignals, the VOI is completely enclosedwithin the brain and positioned atthe center of the phase-encoded fieldof view (FOV), which is large enoughto prevent wraparound artifact. Forlesions that are cortically based ornear the skull base, the VOI can berotated, and up to 8 outer volumesaturation (OVS) slabs can be placedover the skull (Fig. 1). A typical VOIconsists of an 80 mm x 80 mm regionplaced within a 160 mm x 160 mmFOV on a 10 mm to 20 mm slice.

Introduction

Advanced MRI techniques, such as MR spectroscopy, diffusion andperfusion MR imaging can giveimportant in vivo physiological andmetabolic information, complemen-ting morphologic findings fromconventional MRI in the clinicalsetting. Combining perfusion MRIand MR spectroscopy will help insolving difficult cases and increaseconfidence in making a diagnosis.

Techniques

Brain tumor MR Imaging protocol: typical parameters for brain tumorimaging at 1.5 Tesla are shown inTable 1. Perfusion MRI exams areperformed with TR/TE = 1,000 ms/54 ms; field of view 210 mm x 210mm; section thickness 3-8 mm(typically 5 mm); matrix 128 x 128;in-plane voxel size 1.8 x 1.8 mm;intersection gap 0%-30%; flip angle30°; signal bandwidth 1,470 Hz/pixel.Ten slices are usually obtained tocover the entire lesion volume asidentified on T2-weighted images. Aseries of 60 multi-slice acquisitionsare acquired at 1-second intervals.The combination of a 1,000-msecrepetition time and a 30° flip angleensures that T1 effects are minimized.The first 10 acquisitions are per-formed prior to the contrast agentinjection to establish a pre-contrastbaseline. At the 10th acquisition,gadopentetate dimeglumine (0.1 mmol/kg) is injected with a


CLINICALMRS BRAIN

Figure 1

Figure 2 A 52-year-old woman with a histologically confirmed grade II/IVglioma. A: Axial T1-weighted image post-gadolinium shows a lesion in the righttemporo-parietal region with low signal and minimal enhancement. B: AxialFLAIR image shows increase in T2 signal within the lesion with minimal edema.C: Gradient-echo axial perfusion MRI with rCBV* color overlay map, shows alow rCBV* of 1.70 in keeping with a low-grade glioma. There is a thin rim ofslightly increased perfusion at the margin of the lesion. D: Cho color overlaymetabolite map showing the region of increased Cho, which sometimes doesnot correspond to the region of highest rCBV* seen on the perfusion MRI studyin C. This may be due to the fact that MRSI and perfusion MRI are measuringdifferent parameters of tumoral activity. E: Spectral map shows regions of Choelevation only within the region of abnormal signal indicating a lack of tumorinfiltration in the peritumoral region in this grade II lesion. F: Spectrum (TE 144ms) showing Cho elevation with respect to Cr and NAA within the tumor. Datawere acquired on a 1.5T MAGNETOM Symphony.

Sequence TR TE Flip Angle Acq/NEX Thickness No. Slices Matrix FOV Acq. time(ms) (ms) /TI (mm) (mm) (min.sec)

Scout/Localizer 15 6 NA 1 8 3 256 280 0.19

Axial T1 600 14 90 2 5 20 256 210 3.36

Axial FLAIR 9000 110 180/2500 1 5 20 256 210 3.56

Axial T2 3400 119 180 1 5 20 256 210 1.36

1. Dual Echo/PD 3400 16 180 1 5 26 256 256 7.59#

Diffusion/ADC 3400 95 NA 3 5 20 128 210 1.15

2. DTI* 4000 95 NA 4 5 20 128 210 1.56

PERFUSION MRI 1000 54 30 degrees 60 (1/s) 3 to 8 10 128 210 1

Post Gd T1 600 14 90 1 5 20 256 210 3.36

MRSI 1500 144 90 2D CSI 3 10 1 16 x 16 160 6.0530 90 3D CSI 1 10 8 12 x 12 160 7.53

3. MP Rage 1100 4.38 15 1 0.9 192 256 230 3.33

Note: Total imaging time is approximately 30 mins. The optional sequences are:

1. Dual echo T2 – Proton density for patients in a stereotactic headframe, acquired instead ofthe FLAIR. Larger field of view (FOV) andsquare FOV (256x256) are used to match thesoftware in the operating room for stereotacticbiopsy or resection. No angulation.

2. Diffusion Tensor Imaging (DTI in 12 directions).

3. MPRage – reconstructed in the axial, coronaland sagittal planes. TI denotes Inversion Time.

Table 1 Brain tumor imaging protocol.

Cr

Cr2

Cho

A C E

B D F

NAA

*WIP: The information about this product ispreliminary. The product is under developmentand is not commercially available in the US andits future availability cannot be ensured.


CLINICALMRS BRAIN

Figure 3 MRSI and perfusion MRI images from a patient demonstra-ting recurrent high-grade glioma within the left parietal region. Imagesillustrate typical features of spectra at different echo times (TE). A: Thefirst column of images shows axial FLAIR images with the voxel positioncorresponding to the spectra in the same row. Short TE (30 ms), inter-mediate TE (144 ms), and long TE (288 ms) spectra are displayed in thesecond, third, and fourth column respectively. In the fifth columncorresponding rCBV* color overlay maps, (B), and (C), of the recurrentglioma demonstrate evidence of increased rCBV* in regions of signifi-cant Cho elevation. The 1st and 2nd rows of (A) represent spectra froma voxel within the recurrent tumor demonstrating marked elevation inCho relative to Cr and NAA. There is also a lactate peak that is invertedat the intermediate TE of 144 ms and upright at TEs of 30 and 288 ms.Note that myo-inositol, Glx (Glutamine/Glutamate) and lipid peaks aremore readily appreciated at the short TE (2nd row, 30 ms). There is anelevated Glx complex of peaks (2.05-2.5 ppm). (In addition to the Glx,there may be a very small NAA peak that disappears at long TE becauseof T2 decay). The 3rd row displays spectra from a voxel within normalappearing contralateral brain. Note the decrease in signal (from meta-bolites) relative to baseline noise at longer echo time (288 ms) due tothe T2 decay of signal from metabolites at longer echo time. The 4throw displays a voxel in the ventricles, where the lack of metaboliteswithin CSF results in a noisy spectrum, seen at all echo times (30, 144and 288 ms). Data were acquired on a 1.5T MAGNETOM Symphony.

Voxel position TE = 30 ms TE = 144 ms TE = 288 ms


CLINICALMRS BRAIN

Figure 4 A 70-year-old woman with a histologically confirmed gradeIII/IV glioma. A: Post-gadolinium Axial T1-weighted image shows a lesion in the right thalamic region with heterogeneous peripheralcontrast enhancement and a central cystic/necrotic region. B: AxialFLAIR image shows increase in T2 signal within the lesion with moderatesurrounding edema. The patient also has hydrocephalus and trans-ependymal edema around the ventricles. C: Gradient-echo axial per-fusion MRI with rCBV* color overlay map, shows a high rCBV* of 3.70 in keeping with a high-grade glioma. There is a thick rind of markedincreased perfusion. D: Cho color overlay metabolite map showing theregion of increased Cho, which again does not correspond exactly to the region of highest rCBV* seen on the perfusion MRI study in C. E:Spectral map shows regions of Cho elevation within the region ofabnormal signal as well as tumor infiltration beyond the region ofenhancement, anteriorly and potentially across the midline into the leftthalamus. The central areas do not demonstrate substantial amount oflipids/lactate as compared to a grade IV lesion. F: Spectrum (TE 144 ms)from the peritumoral region showing marked Cho elevation withrespect to Cr and NAA indicating tumor infiltration. Data were acquiredon a 1.5T MAGNETOM Symphony.

*WIP: The information about this product is preliminary. The product is under developmentand is not commercially available in the US and its future availability cannot be ensured.

Cr

Cr2

Cho

A C E

B D F

NAA


CLINICALMRS BRAIN

1.02±0.42 [3], in (Fig. 6). NormalCho, elevated myo-Inositol, anddecreased NAA have been demon-strated with MRS in gliomatosiscerebri [4] (Fig. 6). The combinationof these spectroscopy findings andreduced perfusion suggest thatgliomatosis cerebri can be differen-tiated from high-grade multicentricglioma.

Metastatic Neoplasms A solitary brain metastases may beindistinguishable from a primaryglioma by conventional MRI imaging.The pathophysiology of the peri-tumoral region of gliomas and metas-tases is well known by neurosurgeonsand neuro-pathologists. High-gradegliomas are known to be infiltratingtumors, with tumoral tissue infiltra-ting along vascular channels, whereasin metastases, the peri-tumoral regioncontains no infiltrating tumor cells orvascular endothelial proliferation andis almost purely vasogenic edema [5-7]. In the differentiation of gliomafrom metastases, finding high Choand high rCBV* in the peri-tumoralregion of a lesion (Figs. 7-8) is morelikely to represent glioma rather thanmetastases [7].

Non-Neoplastic Mimics –Cerebrovascular Injury The clinical differentiation between a cerebral infarct and glioma isusually straightforward, with acutecerebrovascular events presentingwith notable neurologic symptomsand signs. Diffusion-weighted imagingis also extremely sensitive and timeefficient in demonstrating ischemiaand infarction. However, in someclinical instances where the clinicalhistory and diffusion-weightedimaging is somewhat confusing (Fig. 9),MRSI and perfusion MRI can be usedto differentiate a surgical lesion(tumor) from a non-surgical lesion

A C E

B D F

CrCr2

ChoNAA

Figure 5 A 60-year-old man with a histologically confirmed grade IV/IVglioma. A: Post-gadolinium axial T1-weighted image (motion degraded)shows a lesion in the right temporal region with heterogeneous peripheralcontrast enhancement and central necrosis. B: Axial FLAIR image showsincrease in T2 signal within the lesion with marked surrounding edema/tu-mor infiltration. There is also infiltration into the right occipital region andright midbrain. C: Gradient-echo axial perfusion MRI with rCBV* color over-lay map, shows a high rCBV* of 8.90 in keeping with a grade IV glioma. Thereis increased perfusion around the periphery of the lesion with decreasedperfusion within the necrotic center. D: Cho color overlay metabolite mapshowing the region of increased Cho which corresponds to the regions ofhigh rCBV* seen on the perfusion MRI study in C. E: Spectral map showsregions of lipid within the region of abnormal signal indicating necrosis in a glioblastoma multiforme. F: Spectrum (TE 30 ms) showing markedlipids elevation (primarily aliphatic methylene groups [-CH2-] of fatty acids at 1.2-1.4 ppm). Data were acquired on a 1.5T MAGNETOM Symphony.


CLINICALMRS BRAIN


A

B C

Figure 6 A: Axial T2-weighted image shows extensive hyperintensityinvolving the right hemispheric white matter. Contiguous involvement of atleast two lobes of the brain is characteristic of gliomatosis cerebri. B: Gradientecho perfusion MRI with rCBV* color overlay demonstrating reduced per-fusion. Mean rCBV* was 1.02. C: Multi-TE MRSI of gliomatosis cerebri. Thefirst column shows axial T2-weighted images with the voxel positioncorresponding to the spectra in the same row. Short TE (30 ms), intermedi-ate TE (144 ms), and long TE (288 ms) spectra are displayed in the second,third, and fourth columns respectively. The 1st row demonstrates spectrafrom within the tumor, at short TE, there is elevation in myo-inositol anddecrease in NAA without appreciable increase in Cho seen at all echo times(30, 144 and 288 ms). The 2nd and 3rd row demonstrate normal spectrafrom within the normal appearing contralateral brain and the relativelyunaffected right thalamus respectively. The 4th row demonstrates eleva-tion in myo-inositol at short TE (30 ms) and some elevation in Cho withinthe tumor. Myo-inositol elevation in the setting of normal Cho has beendescribed in gliomatosis cerebri (4). Increased Cho/Cr and Cho/NAA havealso been described with longer TE’s (3). Data were acquired on a 1.5T MAGNETOM Symphony.


CLINICALMRS BRAIN

(stroke). Strokes generally demon-strate increased mean transit time(MTT), and decreased CBF [8-13].Strokes may demonstrate elevated ordecreased rCBV*. In contrast, tumorsgenerally demonstrate increasedrCBV* and CBF*. Review of the MTT,CBF* and rCBV* measurements and color overlay maps will not onlydemonstrate substantial reduction in perfusion in a stroke comparedwith a tumor but it will conform to a vascular territory (Fig. 9).

An initial inspection of spectros-copy data obtained in a stroke willdemonstrate Cho/Cr elevation due tocell membrane destruction anddemyelination, as well as decrease inNAA from neuronal and axonaldamage. Lactate is also demonstratedfor up to 6 weeks and sometimesbeyond, initially from anaerobicglycolysis, then later from macrophageactivity and persistent ischemia [14].However, the major differencebetween a stroke and tumor can bestbe appreciated by comparing aspectrum from within the stroke to aspectrum from contralateral healthytissue, which can be done with amulti-voxel MRSI acquisition. Theprimary difference is a reduction inall metabolites in stroke comparedwith the contralateral normal brain.Cr, as an energy marker, is particularlyreduced in stroke. Even though theremay be Cho/Cr elevation and NAA/Crdecrease, when comparing the Choto normal contralateral Cho(n), therewill be a reduction in the Cho/Cho(n)ratio. This demonstrates the value ofdirectly comparing abnormal withnormal metabolite levels and measu-ring Cho/Cho(n) Cho/Cr(n) and NAA/NAA(n) ratios which may increasethe specificity somewhat in charac-terizing mass lesions in the clinicalsetting when absolute quantification,relative to an external standardcannot be made.

Figure 7 A glioblastoma multiforme. A: Axial T1-weighted image post-gadolinium shows a peripherally enhancing mass with heterogeneous signalintensity and central necrosis in the left parietal region. B: Axial T2-weightedimage demonstrates moderate surrounding T2 signal abnormality, which is likely to represent infiltrating tumor and edema. C: Cho metabolite coloroverlay demonstrates elevation in Cho in and around the region of enhance-ment and T2 signal abnormality. D: MRSI (TE 144 ms) spectral map confirmsthe increase in Cho within the lesion as well as within the abnormal T2 signalanteriorly. E: Spectrum (TE 144 ms) from within the anterior peri-tumoralregion demonstrating increase in Cho/Cr and in Cho/NAA from tumoralinfiltration of adjacent peri-tumoral tissues. F: Gradient-echo axial perfusionMRI with rCBV* color overlay show increase in vascularity within the enhancingtumor as well as in the peritumoral region anteriorly. These MRSI and per-fusion MRI findings in the peritumoral region help to differentiate an infiltra-ting primary high-grade glioma from a metastasis. Data were acquired on a 3T MAGNETOM Allegra.

A C E

B D F

Cr

Cr2

Cho

NAA


CLINICALMRS BRAIN

Figure 8 60 year-old-woman with a metastasis from breast carcinoma. A: Axial T1-weighted image post-gadolinium shows a well defined mass inthe left frontal region. B: Axial T2-weighted image demonstrates negligibleedema surrounding this lesion. C: Cho metabolite color overlay demonstratesCho/Cr elevation within the tumor and in voxels where there is partial volumeaveraging with the tumor. D: MRSI (TE 144 ms) spectral map demonstratesno elevation in Cho within voxels in the peri-tumoral region of the lesion,indicating a non-infiltrating lesion like a metastasis. E: Spectrum (TE 144 ms)from a voxel in the peri-tumoral region with normal metabolite peaks. F: Gradient-echo axial perfusion MRI with rCBV* color overlay may demon-strate increase in tumor vascularity confined to the enhancing tumor but notwithin the peri-tumoral region. Data were acquired on a 1.5T MAGNETOMSymphony.


A C E

B D F

Cr

Cr2

Cho

NAA


CLINICALMRS BRAIN

Figure 9 45-year-old-woman presenting with a 10-day history of right armand leg weakness: The clinical and final diagnosis was a left middle cerebralartery territory stroke, however conventional MRI, including diffusion weightedimaging was inconclusive. A: Axial T1-weighted image post-gadoliniumdemonstrates abnormal enhancement in the left hemisphere involving grayand white matter. B: Axial FLAIR image demonstrates abnormal T2 signal as well as some mass effect. There is a small amount of edema. C: Diffusion-weighted image (B = 1000) demonstrating some increase in signal. D: TheADC map also shows areas of increased signal, which may represent “T2-shine-through effect” as well as some areas of decrease signal, which mayrepresent true diffusion restriction. These findings on diffusion are sometimesalso seen in very heterogeneous high-grade gliomas. E: Gradient-echo axialperfusion MRI with mean transit time (MTT) color overlay demonstratingprolongation in the MTT throughout the entire middle cerebral artery (MCA)territory in keeping with left MCA ischemia. F: Signal intensity versus timecurves color coded to the 4 ROIs placed on the MTT maps, demonstratingprolongation in MTT in the green, blue and to a lesser extent the yellow ROIs.The red ROI indicates normal MTT in the contralateral hemisphere. The delayin MTT and reduced CBF* in the MCA distribution is in keeping with a stroke.Data were acquired on a 1.5T MAGNETOM Symphony.

A C E

B D F


CLINICALMRS BRAIN

References [ 1 ] Law M, Yang S, Wang H, et al. Glioma Grading: Sensitivity, Specificity andPredictive Value of Perfusion MRI and ProtonSpectroscopic Imaging compared with Conven-tional MR Imaging. AJNR 2003 24: 1989-1998.

[ 2 ] Kleihues P, Cavanee P. WHO Classification of Tumors: Pathology andGenetic of Tumours of the Nervous System. Lyon: IARC Press, 2000.

[ 3 ] Yang S, Wetzel S, Law M, et al. Dynamic contrast-enhanced T2*-weighted MRimaging of gliomatosis cerebri. AJNR Am J Neuroradiol 2002; 23:350-355.

[ 4 ] Saraf-Lavi E, Bowen BC, Pattany PM, et al.Proton MR Spectroscopy of Gliomatosis Cerebri: Case Report of Elevated Myoinositolwith Normal Choline Levels. AJNR 2003; 24:946-951.

[ 5 ] Burger PC. Classification, grading, andpatterns of spread of malignant gliomas. Neurosurgical topics: malignant cerebralglioma. America Association of NeurologicalSurgeons. Park Ridge, Ill, 1990; 3-17.

[ 6 ] Burger PC, Vogel FS, Green SB, et al.Glioblastoma multiforme and anaplasticastrocytoma: Pathologic criteria and prognosticimplications. Cancer 1985; 56:1106-1111.

[ 7 ] Law M, Cha S, Knopp EA, et al. High-Grade Gliomas and Solitary Metastases: Differentiation by Using Perfusion and ProtonSpectroscopic MR Imaging. Radiology 2002; 222:715-721.

[ 8 ] Karonen J, Liu Y, Vanninen R, et al.Combined Perfusion- and Diffusion-weightedMR Imaging in Acute Ischemic Stroke during the 1st Week: A Longitudinal Study. Radiology 2000; 217:886-894.

[ 9 ] Sorensen AG, Copen WA, Ostergaard L, etal. Hyperacute stroke: simultaneous measure-ment of relative cerebral blood volume, relativecerebral blood flow, and mean transit time.Radiology 1999; 210:519-527

[ 10 ] Sorensen AG, Buonanno FS, Gonzalez RG, et al. Hyperacute stroke: evaluation with combinedmultisection diffusion-weighted and hemody-namically weighted echo-planar MR imaging.Radiology 1996; 199:391-401.

[ 11 ] Hatazawa J, Shimosegawa E, Toyoshima H, et al. Cerebral blood volume in acute brain infarction:A combined study with dynamic susceptibilitycontrast MRI and 99 mTc-HMPAO-SPECT. Stroke 1999; 30:800-806.

[ 12 ] Crosby D, Simonson T, Fisher D, et al.Echo-planar MR Imaging: correlation ofintracranial perfusion-sensitive imaging withcerebral angiography. Society of Magnetic Resonance 1994; 277.

[ 13 ] Saunders DE. MR spectroscopy in stroke.Br Med Bull 2000; 56:334-345.

Figure 10 45-year-old-woman with a left MCA stroke. A: MRSI (TE 144 ms)spectral map demonstrating an increase in Cho/Cr within the abnormalvoxels in the left hemisphere, however, there is an overall decrease in meta-bolites when compared with the right hemisphere. B: The Cho/Cr metaboliteratio color overlay map shows increase in Cho/Cr, which can be misinter-preted for a neoplastic process. C: The Cho metabolite color overlay demon-strates a decrease in Cho (but NOT Cho/Cr) when compared with the righthemisphere. D: MRSI (TE 144 ms) showing a spectrum from the normal righthemisphere. Note is made of the vertical scale and the normal Cho level [Cho(n)], which is used as a control for comparing the abnormal side. E: A spectrumfrom the abnormal left side displayed on the same scale as the spectrum inD. It shows a decrease in all metabolites including choline, which indicatesan ischemic rather than neoplastic process. F: The same spectrum from Escaled to the highest peak in the spectrum i.e choline may be misinterpretedas a neoplastic process. This display illustrates the pitfall of reading a spec-trum without a contralateral control. Finding substantial reduction in Cr, Choand NAA by comparing a spectrum with the normal contralateral control canincrease the specificity and exclude a neoplastic process in favor of infarct inthis case. Data were acquired on a 1.5T MAGNETOM Symphony.


A C

E

B

D F

the same parameters, and repeatedonce without a pause. The subjectwas then brought out of the magnetfor a short break, after which he wasrepositioned in the magnet andscanned again. This process wasrepeated 3 times yielding a total 6scans.

The 2D CSI parameters were:(sequence CSI_se_135), FOV =160x160x15 mm3, number of voxels16x16x1, resolution per voxel =10x10x15 mm3, TR=1500,TE=135,NA=4, and TA=7’06”. Other settingsincluded a Hamming filter, numberof sampling points NP =1024, BW =1kHz, preparatory scans = 4, phaseencoding = weighted, WS BW = 35Hz,delta frequency = -2.7ppm.

Results

The slice of interest used for the dataevaluations reported here is shown inFigure 1. The region of interest with6 x 6 voxels is shown in Figure 1aalong with the coronal and sagittalviews. A single voxel spectrum isshown in Figure 1b and a CSI exam-ple for each voxel is shown in Figure1c. Integral values of metabolitepeaks in each voxel were used in theevaluation.

Average standard deviationover all voxels: Assuming that thestandard deviation over the 8 measu-rements is the same for all voxels, itis reasonable to quote the meanstandard deviation over all 36 voxels(column 1 of table 1). A histogram ofthe standard deviations shows thatthis is in fact reasonable.

Precision of the mean of allvoxels from scan to scan: The nextstep was to compare the variation inthe means over the 8 repetitions.Ideally, one could compare thevalues in each voxel but that wouldrequire perfect spatial match of the36 voxels. Since the voxel to voxelvariation is seen to be rather large, a


CLINICALMRS BRAIN

Jiani Hu, Yimin Shen, Yang Xuan and E. Mark Haacke

Wayne State University, Magnetic Resonance ResearchFacility

MR Spectroscopy Precision andRepeatability: Evaluation of Brain CSI Data

Introduction

With the introduction of any newsystem, the question of hardwareand software performance comesinto play. Our goal in this study wasto determine the precision of the CSImeasurements offered by the Sonataand our ability to reproduce theseresults when the data are re-acquiredon separate days and separate ima-ging scenarios.

CSI is used for monitoring theeffect of therapy on certain patholo-gies such as tumor and multiplesclerosis. A knowledge of the varioussources of error that affect CSI resultsis critical for drawing conclusionsfrom repeated measurements on thesame subject in a longitudinal study.Similarly, the natural variability indifferent voxels in the brain may bevery different from the noise levelsapparent within a given voxel. There-fore to draw conclusions as to what isnormal and what is not within agiven voxel requires knowledge ofthis variation.

Methods

A young healthy 22 year old malevolunteer was scanned first in thespring of 2003. A special foam headconstraint was used in order to keepthe head fairly still. This eliminateserrors from head motion that mayotherwise obscure statistical errorsinherent to the sequence and soft-ware associated with CSI. While inthe magnet, the subject was scanned8 consecutive times with no pausebetween scans. A month later, thesame subject was positioned in themagnet without the foam headhol-der. Using an anatomical landmarkon the forehead an attempt wasmade to reproduce the same headpositioning used earlier. A CSI mea-surement was acquired at the sameanatomical slice position and with

Abstract

Our goal in this study was to determi-ne the precision of the CSI (Chemicalshift Imaging) measurements offeredby the Sonata and our ability toreproduce these results when thedata are re-acquired on separate daysand separate imaging scenarios.Variations of the CSI measurementswithin the brain are also reported.Results of this work suggest that theprecision of the integral values for agiven voxel is on the order of 6% forNAA, 10% for Cre and 11% for Cho.The reproducibility is found to bealmost identical implying that theresults from one day to the next arein good agreement with each otherstatistically. Variation in metabolitecontent throughout the brain givesvoxel variability on the order of 10%for NAA, 12% for Cre and 16% forCho. The ratios of these metaboliteswill, of course, have appropriatelylarger errors.


CLINICALMRS BRAIN

Table 1 An estimate of the error in the means from 8 scans.

small offset or change in partialvolume effect could make a bigdifference in this test of reproducibili-ty. To avoid this problem, we compa-red only the means over all 36 vo-xels. The global mean (µg column 2 oftable 1) of each metabolite was thenfound over all 36 voxels. As might beexpected, the global mean had a verylow standard deviation (column 4 oftable 1).

Variance from voxel to voxelwithin the brain: Finally, a compari-son of the spatial variation in meta-bolite concentration was made over

Metabolite or µvs Mean of µg Mean 100 µvs/µg Sd of the µspat 100 µspat/µg

ratio of all the Sd of over all global mean Spatial Sd metabolites all 36 single 36 voxels over 8 scans of the means

voxels over all 36 voxels

naa 0.18 3.12 6% 0.02 0.31 10%

cr 0.17 1.64 10% 0.04 0.20 12%

cho 0.18 1.67 11% 0.07 0.26 16%

naa/cr 0.24 1.93 12% 0.06 0.33 17%

cho/cr 0.16 1.04 15% 0.05 0.23 22%

Table 2 An estimate of the error in the means from 6 scans.

Metabolite or µvr Mean of µg Mean 100 µvr/µg Sd of the µspat 100 µspat/µg

ratio of all the Sd of of integral global mean Spatial Sd metabolites all 36 single 36 voxels over 6 scans of the means

voxels value over all over all 36 voxels 36 voxels

naa 0.21 3.16 7% 0.07 0.32 10%

cr 0.19 1.68 11% 0.03 0.22 13%

cho 0.21 1.74 12% 0.10 0.26 15%

naa/cr 0.25 1.92 13% 0.03 0.35 18%

cho/cr 0.17 1.05 16% 0.05 0.21 20%

the 36 voxels in this particular slice.This of course could vary from loca-tion to location in the brain. Thespatial variance is much larger thanthe precision (µspat column 5 of table1).

Precision of the mean overdifferent sittings: All 6 cases wereincluded in the analysis. Despite thefact that these scans were acquired ondifferent days, the spatial and tissuevariations were only slightly largerthan in the single sitting case (seetable 2).

Discussion and ConclusionsWe have tested the same day precision(data set 1, table 1) and reproduci-bility from day to day (data set 2,table 2) of the 2D CSI approachcurrently available on the SiemensSonata scanner. It is encouragingthat both the means and standarddeviations are statistically consistentwith each other. In order to be ableto have the confidence in the lowerstandard deviations, it is critical tovery carefully reposition the subject.In this case, variations will fall withinthe expected precision values quoted

Figure 1b

Figure 1a


CLINICALMRS BRAIN


CLINICALMRS BRAIN

in table 1. In these tables we quotethe standard deviation. However,since we used 8 scans in table 1, thestandard error of the mean will bethe quoted standard error divided bysqrt(7).

Clearly, to determine that anappreciable change has occurredover time in a disease or after treat-ment one would look for changes ofat least two standard deviations.Unfortunately, reducing the noiselevels to less than 3% for NAA and 5%for Cho and Cr requires 4 scans orroughly 30 minutes. This is certainlynot impossible, but the need for itshould be determined based on thesize of the change being sought.However, if this is done, it is recom-mended that the scans be run sepa-rately as we did above so that ifmotion does occur for a given scanthat at least one of the scans remainsusable.

In conclusion, with a 7 minutescan, it is possible to obtain estimatesfor NAA, Cre and Cho with errors onthe order of 6%, 10% and 11%,respectively (see column 3 in table 1and 2). With careful patient positio-ning, this precision can be replicatedeven when imaging the same personon another day.

Editor’s note: Please note that the new featuresPhoenix supporting MRS protocols,and auto-align, both available with syngo MR 2004A/V, will furthersupport you to increase thereproducibility of your results.

Figure 1c

functional information in the opera-ting room, leading to so-called func-tional neuronavigation. In conven-tional MRI it is often difficult todelineate the heterogeneous structureof lesions, especially of gliomas. Even the current method of choice,contrast-enhanced MRI as a techniquefor visualizing regions where theblood-brain barrier is damaged, is nottumor specific and can result inambiguous or misleading results [1,2]. Establishing the position and thesize of the border zone betweentumor and normal brain tissue is oneof the major problems in therapyplanning. Proton magnetic resonan-ce spectroscopic imaging (1H-MRSI)allows noninvasive measurements of the concentration and spatialdistribution of brain metabolites andtherefore may provide biochemicalinformation in vivo, useful in distin-guishing pathologic from normalbrain areas [3]. Brain tumors showincreased levels of choline-contai-ning compounds (Cho) and a reduc-tion in N-acetyl-aspartate (NAA) andcreatine (Cr). Cho is thought to be a marker for increased membraneturnover or higher cellular density[4]. NAA is mainly contained withinneurons [5] and Cr is a marker forenergy metabolism [6]. The range ofCho increase and NAA decrease iscompatible with the range of tumorinfiltration [1, 7]. Metabolite maps of NAA and Cho allow the differen-tiation of areas of necrosis, solidtumor and varying degrees of tumorinfiltration and tissue edema.

Materials and Methods

3D-MRI and 1H-MRSI:All MR studies were performed on a MAGNETOM Sonata 1.5 Tesla(Siemens Medical Solutions, Erlangen,Germany) equipped with the stan-dard head coil. Tumor patients – allwith supratentorial gliomas (WHO


CLINICALMRS BRAIN

1,2Andreas Stadlbauer, MSc, Erlangen2Stephan Gruber, Ph.D. 2Ewald Moser, Ph.D., Vienna1Christopher Nimsky, M.D., Erlangen1Peter Grummich, Ph.D. 1Rudolf Fahlbusch, M.D., Erlangen1Oliver Ganslandt, M.D.

1Department of Neurosurgery,Neurocenter, University of Erlangen-Nuernberg,Erlangen, Germany

2Centre of Excellence for High-FieldMR, Medical University of Vienna,Vienna, Austria

1H-MRSI Guided Surgery of Brain Tumors

Abstract

Proton magnetic resonance spectro-scopic imaging (1H-MRSI) is a non-invasive tool to measure the spatialdistribution of brain metabolites. This technique may therefore providebiochemical information in vivo,useful in distinguishing pathologicfrom normal brain areas. However, itis often difficult to delineate thetumor borders of high and low gradegliomas in conventional T1- and T2-weighted magnetic resonance ima-ging (MRI). We have developed amethod to intraoperatively investigatepathologic changes in the spatialdistribution of choline-containingcompounds (Cho), total creatine (Cr)and N-acetyl-aspartate (NAA) in braintumors combining 1H-MRSI withframeless sterotaxy. Metabolic mapswere calculated and segmentation of the tumors was performed. Spec-troscopic images of the segmentedtumor were matched onto an ana-tomical 3D-MRI set and used sub-sequently for neurosurgical planning.Display of spectroscopic informationin the navigation microscope wasperformed during surgery leading to1H-MRSI guided neuronavigation. We conclude that our method maypresent another step towards intra-operative identification of tumorborder zones based on the metabolicchanges due to tumor infiltration.

Introduction

The main goal in neurosurgery oflesions is to achieve tumor resectionas complete as possible while pre-serving normal brain tissue and func-tion. Information about the localiza-tion and spatial extent of the lesionshould be available before andduring the surgical procedure. Frame-less stereotactic methods in neuro-surgery (neuronavigation) have beenused as a platform to integrate


CLINICALMRS BRAIN

Figure 1 Anatomicalimages (T2-weighted)overlaid with metabolicmaps of a patient with a gliomblastoma multiforme (GBM). a: Cho map; b: NAA map; c: map of the Cho/NAAratios with selected“healthy region” (redrectangle); and d: map of the segmentedtumor. Spectra of (1)contralateral normalbrain, (2) the tumorborder, (3) the transitionto tumor center, and (4) the tumor centeraccording to the markedpositions in a to d.

grade II to IV) – were examined usinghigh-resolution 1H-MRSI. In eachMRSI session a localization scan andan axial spin echo (SE) sequence (T1-weighted) were acquired forMRSI excitation volume location. TheSE sequence was used for matchingspectroscopic images to an anatomicthree-dimensional magnetic reso-nance image set. The parameterswere TR/TE 500/15 ms, 256x256matrix size, 16x16 cm FOV, 20 sliceswith a distance factor of 0% and a slice thickness of 2 mm. The MRSIsequence, which used PRESS (Point-RESolved Spectroscopy) volumepreselection, was performed immedi-ately afterwards. Water suppressionwas achieved using three CHESS(CHEmical Shift Selective) pulsesprior to the PRESS excitation. TheMRSI parameters were TR/TE1600/135 ms, 24x24 circular phase-encoding scheme across a 16x16 cmFOV, slice thickness 10 mm, 50%Hamming-filter and 2 NEX, spectralwidth 1000 Hz and 1024 complexpoints acquisition size. The totalspectroscopic data acquisition timewas less than 13 minutes.

The MRSI slab was aligned preci-sely to a selected SE slice by copy andpaste of the image position. Due tothe same FOV of the SE and the MRSIexperiment, and with the assump-tion of negligible head motionbetween the time of the MRI andMRSI acquisition, direct correlation ofthe data of the MRSI slab (10 mm thick)with five slices (each 2 mm thick) ofthe anatomical MRI was achieved.Immobilization of the patients’ headswas achieved by fixation in a head-rest.

In a single session a three-dimen-sional anatomic magnetization pre-pared rapid acquisition gradient echo(MPRAGE) sequence was performedwith the following parameters: TR/TE2020/4.38 ms, 25x25 cm FOV, 1 mmisotropic and 160 slices. For registra-

tion in the neuronavigation systembetween 6 and 8 adhesive skin fiducials were placed in a scatteredpattern on the head surface prior toimaging.

MRSI Data Analysis:The peak areas for Cho, Cr and NAAwere calculated by integration overthe frequency range of 3.34-3.14ppm, 3.14-2.94 ppm and 2.22-1.82ppm, respectively (see spectra #1 to#4 in Fig. 1). Smooth linear inter-polation to a 256x256 matrix resultedin metabolic maps. Cho and NAAimages (Figs. 1a and b) were used tocalculate a map of Cho/NAA ratios(Fig. 1c). Tumor segmentationprocedure was developed on theassumptions that (i) the values for

Cho/NAA in normal brain follow aGaussian distribution, and (ii) thosefor the tumor, including the borderzone, are significantly increased. For segmentation we determined a‘healthy region’, unaffected by predo-minantly white matter on the contra-lateral side and at sufficient distancefrom the lesion according to theCho/NAA ratio map (red rectangle inFig. 1c). The mean and the standarddeviation (SD) of the Cho/NAA ratioswere calculated for this selectedregion and all Cho/NAA values in thewhole map less than the mean+3SDwere set to zero. This leads to theelimination of all normal brain areasin the Cho/NAA map and hence tothe segmentation of the tumor (Fig. 1d) [8].

a 1

b

c d

2

3

4

Figure 2 Image fusion of metabolic maps (MRI/MRSI hybrid data set, inorange) to a 3D-MRI dataset (in green) of a patient with an oligodendrogliom(WHO grade III). The result is a 3D-MRI consisting of anatomical and metabolicinformation for surgical planning. A: Axial, B: sagittal and C: coronal recon-struction of the 3D data set after image fusion, respectively. (Note: The small light green rectangles below the MRSI (orange) in (B): are results of a MEG examination (language stimulation).

Figure 3 1H-MRSI guided neurosurgical planning. A: and B: reconstruction of a sagittal and coronal view from the stereotactic 3Ddata set (same patient as in Fig. 2). The pink line represents the segmentation of the tumor margin with help of MRSI which was done manually by a neuro-surgeon. (Note: The orange line presents the region of interest based on theMEG examination, same as Fig. 2b.) C: 3 D reconstructed tumor resection plan.The target volume was the pink volume. (Note: The orange volume is the 3D reconstructed result of the MEG experiment.)

DiscussionWe present a technique for the inte-gration of high-resolution 1H-MRSIinformation into the neurosurgicaloperating room. We used a combinedimaging data set consisting of con-ventional MRI slices and segmentedmetabolic maps of the MRSI slice, aso called MRI/MRSI hybrid data set,for merging biochemical informationin a global anatomical 3D MRI.Drawing ROIs based on high-resolu-tion metabolic images allows intra-operative visualization of the spectros-copic information displayed through


CLINICALMRS BRAIN

Tumor patients underwentsurgery with functional neuronaviga-tion after integration of 1H-MRSI. The projection of segmented MRSIdata into the operating viewing fieldallowed easy identification of thetumor border based on biochemicalinformation (Fig. 4a to c).

The total time for performing thisprocedure was about one hour. Thiswas divided into 20 minutes for con-ventional MRI (SE sequence) and MRSIdata acquisition, 30 minutes for MRSIdata analysis, and 10 minutes for ob-taining the MRI/MRSI hybrid data set.

MRI/MRSI hybrid dataset andframeless stereotaxy: A so-called MRI/MRSI hybrid datasetconsists of both anatomical andmetabolic information. Replacing ofraw images of the five correlatedanatomical slices in the SE data set bythe corresponding Cho/NAA ratiomap of the segmented tumor resultsin this hybrid data set [9]. It wastransferred to a frameless stereotacticsystem (VectorVisionSky, BrainLab,Heimstetten, Germany) via fastethernet by using the DICOM 3protocol (Digital Imaging and Com-munications in Medicine) and imagefusion to a 3D MRI dataset wasperformed (Fig. 2a to c). Regions ofinterest (ROIs) were drawn with thehelp of MRSI (Figs. 3a and b). Thenavigation microscope (NC4-Multi-vision, Zeiss, Oberkochen, Germany)in combination with the ceilingmounted navigation system, enablesMRSI guided surgery of brain tumors.

Results

MRSI data analysis, including thecalculation of metabolic maps andsegmentation, as well as the integra-tion of these spectroscopic results in functional neuronavigation wassuccessfully performed in neuro-surgical treatment planning. Fig.1shows the characteristic change inspectral patterns, representingbiochemical information, startingfrom normal (white matter) tissue(spectrum #1: NAA > Cr and Cho, Cr > Cho) via tumor border (spectrum#2: NAA approx. Cho or Cr, Cho > Cr),tumor (spectrum #3: NAA < Cr orCho) and tumor center (spectrum #4:NAA approx. 0, Cho >> Cr). Metabolic maps, in particular forCho/NAA (Fig. 1c and d), clearlyallows visualization and segmen-tation of tumor as well as identifica-tion of tumor center and border zones.

a b c

a b c


CLINICALMRS BRAIN

the eyepieces of the microscope,leading to MRSI guided tumor resec-tion.

Only few studies so far used MRSIto support biopsy target delineation[1, 7, 10-12] or radiation therapytreatment planning [12, 13]. None ofthese studies integrated MRSI data in a neuronavigation system andperformed intraoperative visualiza-tion of MRSI data through imageinjection. Preul et al. [14] achievedintegration of MRSI data (metabolicmaps of Cho) of two patients into animage-guided, frameless stereotactic

system by computing a transforma-tion between the MRSI-space and theglobal MRI-space using the targetingvolume acquired immediately priorto MRSI acquisition. By this approachthey overcame the fact that MRSIlacks detailed structural information.Our strategy for merging MRSI datato a global 3D MRI dataset was thefull accurate integration of metabolicimages with co-registered anatomicalimages (MRI/MRSI hybrid dataset)resulting in a dataset consisting ofboth, anatomical and biochemicalinformation.

In conclusion, integration ofspectroscopic information obtainedby high-resolution 1H-MRSI withframeless stereotaxy may be usefulin glioma resection leading to animproved delineation of the tumorinfiltration zone.

Acknowledgement

We would like to thank the MagneticResonance Spectroscopy Develop-ment Department of Siemens MedicalSystems and especially Stefan Röll for his support. This work was sup-ported by a grant of the DeutscheForschungsgemeinschaft (DFG):GA638/2-1.

References [ 1 ] Dowling C, Bollen AW, Noworolski SM,McDermott MW, Barbaro NM, Day MR, et al.Preoperative proton MR spectroscopic imagingof brain tumors: correlation with histopatho-logic analysis of resection specimens. AJNR Am J Neuroradiol 2001;22(4):604-12.

[ 2 ] Kondziolka D, Lunsford LD, Martinez AJ.Unreliability of contemporary neurodiagnosticimaging in evaluating suspected adult supra-tentorial (low-grade) astrocytoma. J Neurosurg 1993;79(4):533-6.

[ 3 ] Kamada K, Moller M, Saguer M,Ganslandt O, Kaltenhauser M, Kober H, et al. A combined study of tumor-related brainlesions using MEG and proton MR spectroscopicimaging. J Neurol Sci 2001;186(1-2):13-21.

[ 4 ] Michaelis T, Merboldt KD, Bruhn H,Hanicke W, Frahm J. Absolute concentrations ofmetabolites in the adult human brain in vivo:

quantification of localized proton MR spectra.Radiology 1993;187(1):219-27.

[ 5 ] Urenjak J, Williams SR, Gadian DG, Noble M.Proton nuclear magnetic resonance spectros-copy unambiguously identifies different neuralcell types. J Neurosci 1993;13(3):981-9.

[ 6 ] Kemp GJ. Non-invasive methods forstudying brain energy metabolism: what they show and what it means. Dev Neurosci 2000;22(5-6):418-28.

[ 7 ] Croteau D, Scarpace L, Hearshen D,Gutierrez J, Fisher JL, Rock JP, et al. Correlationbetween magnetic resonance spectroscopyimaging and image-guided biopsies:semiquantitative and qualitative histopatho-logical analyses of patients with untreatedglioma. Neurosurgery 2001;49(4):823-9.

[ 8 ] Stadlbauer A, Ganslandt O, Gruber S,Nimsky C, Fahlbusch R, Moser E. Metabolic tumor imaging and absolutequantification of metabolic changes in gliomas. ESMRMB. 2003. Rotterdam. P.194

[ 9 ] Stadlbauer A, Ganslandt O, Gruber S,Nimsky C, Fahlbusch R, Moser E. MR spectroscopy guided brain tumor resection:Integration of MRSI in functional neuro-navigation using a MRI/MRSI hybrid dataset. J Neurosurg. 2003. Rotterdam. P.196

[ 10 ] Hall WA, Martin A, Liu H, Truwit CL.Improving diagnostic yield in brain biopsy:coupling spectroscopic targeting with real-timeneedle placement. J Magn Reson Imaging 2001;13(1):12-5.

[ 11 ] Rock JP, Hearshen D, Scarpace L, Croteau D, Gutierrez J, Fisher JL, et al. Correlations between magnetic resonancespectroscopy and image-guided histopatho-logy, with special attention to radiationnecrosis. Neurosurgery 2002;51(4):912-9; discussion919-20.

[ 12 ] McKnight TR, von dem Bussche MH,Vigneron DB, Lu Y, Berger MS, McDermott MW,et al. Histopathological validation of a three-dimensional magnetic resonance spectroscopyindex as a predictor of tumor presence. J Neurosurg 2002;97(4):794-802.

[ 13 ] Pirzkall A, McKnight TR, Graves EE, Carol MP, Sneed PK, Wara WW, et al. MR-spectroscopy guided target delineation for high-grade gliomas. Int J Radiat Oncol Biol Phys 2001;50(4):915-28.

[ 14 ] Preul MC, Leblanc R, Caramanos Z,Kasrai R, Narayanan S, Arnold DL. Magnetic resonance spectroscopy guided braintumor resection: differentiation betweenrecurrent glioma and radiation change in twodiagnostically difficult cases. Can J Neurol Sci 1998;25(1):13-22.

Figure 4 1H-MRSI guided braintumor resection. a to c: Viewthrough the eyepieces of the navigation microscope. Regions ofinterest (tumor border) as drawn by the neurosurgeon on the basis of the spectroscopic information(and MEG information) are out-lined in green. The dotted linesrepresent the maximum tumordimension as seen from this angle.The green dots are labels for biopsy sampling positions.

a

b

c

examination. These findings are,however, inconstant and in atypicalcases it is difficult to differentiatebetween GC and LGG by conventionalimaging alone. Chemical shift imaging(CSI), also called 2D proton MRspectroscopy, studies the metabolicproperties of brain tissue. We haveshown that the metabolic patterns ofGC and LGG differ strikingly (D Gala-naud et al., J Neurosurg, 2003, 98:269-76). On the one hand, LGG has a classical glial tumor metabolism,associating elevated Cho and mI,reduced NAA and Cr. On the otherhand, GC shows markedly elevatedCr and mI and moderately elevatedCho and reduced NAA. These meta-bolic differences provide the abilityto make a clear-cut differentiationbetween GC and LGG, even in indivi-dual patients.

This data, which is included in ourdatabase of more than 200 tumors,emphasizes the ability of MR spectros-copy and CSI to provide diagnosticand prognostic information for braintumor patients, even when conven-tional MRI and biopsy based pathologyare taken in default.

Scanner Model used: Siemens MAGNETOM Vision plus

Coil used: Head coil


CLINICALMRS BRAIN

Damien Galanaud Ph.D.O.Chinot, F. Nicoli, S. Confort-Gouny,Y. Le Fur, M. Barrie-Attarian, J.P. Ranjeva, S. Fuentes, P. Viout, D. Figarella-Branger, P.J. Cozzone

Centre d’Exploration Métaboliquepar Résonance Magnétique (CEMEREM)Marseilles, France

Case Report: Gliomatosis Cerebri

Images and sequence details

Figure 1Chemical Shift Imaging (CSI) of thepatient (A) and of a control subjectwith low grade glioma (B). Thesequence used was an acquisitionweighted apodized sequence deve-loped in our institution (D. Galanaud et al., MAGMA, 2001, 13:127-33).TR = 1500 ms, TE = 136 ms, 15 mmslice thickness, 6 outer volumesuppression slices are used. Spectraare processed using a proprietarysoftware. NAA: N-Acetyl Aspartate.Cho: Choline containing compounds.Cr: Creatine/phosphocreatine. TheCSI map showing the Cho/Cr ratio is overlaid on a FLAIR image to allowbetter metabolic/anatomic corre-lation.

Discussion

Gliomatosis cerebri (GC) is a diffuseinfiltration of brain parenchyma bytumor cells of glial origin. It is impor-tant to differentiate primitive glioma-tosis cerebri (which has a very poorprognosis) from infiltrating low gradegliomas (LGG), sometimes calledsecondary gliomatosis, which carry a much better prognosis and oftenrespond to chemotherapy and radio-therapy. Conventional pathologydoes not discriminate between thesetwo diseases since it is most com-monly based on small stereotacticbiopsy samples and thus does notprovide a global picture of the lesion.Conventional MR imaging contributesto this differential diagnosis: it hasbeen shown by many authors that GC tends to (i) predominate on thewhite matter and relatively spare the cortex (ii) often invade the basalganglia and/or the brainstem (iii) lack significant focal mass effect andcontrast enhancement at the initial

Patient History A 70-year-old man was referred tothe Neurology Department of La Timone Hospital in Marseillesexhibiting a mental deterioration ofone month duration. Magneticresonance imaging and spectroscopy(including both single voxel andspectroscopic imaging) was per-formed. Based on these radiologicaland metabolic data, gliomatosiscerebri was suspected. A stereotacticbiopsy yielded glial tumor cells,confirming this diagnosis. Chemo-therapy was instituted, but thepatient’s condition rapidly worsenedand he died within a month.


CLINICALMRS BRAIN

Figure 1

Cho

Cr

NAA

ppm ppm

Lesion

Cho/Cr map Cho/Cr map

A B

Cho

Cr

NAA

ppm

Controlateral Cho

Cr

NAA

ppm

Controlateral

Cho

Cr NAA

Lesion

4 3 2 1 0 4 3 2 1 0

4 3 2 1 0 4 3 2 1 0


CLINICALMRS BRAIN

Priv. Doz. Dr. med. F. FellnerDr. rer. biol. hum. C. Fellner

Radiologisches InstitutOberösterreichische Landes-nervenklinik Wagner-JaureggLinz

Spectroscopy in Differential Diagnosis of an Intracranial Mass Lesion

History

The patient was sent to MR withperipheral facial paralysis. MR revealedfluid in the mastoid cells and contrastenhancement in the anatomical area where the peripheral part of thefacial nerve crosses. These findingssupported the diagnosis of infectionin the mastoid cells. The right occipitallobe also revealed a mass lesion of1.5 cm. MR spectroscopy was advisedfor further evaluation of the lesion.

Image Findings

The examination was carried outusing a MAGNETOM Symphony withstandard CP Head Array coil. Forspectroscopy examination, the SE(PRESS) single voxel sequence withecho times of 30, 135 and 270 mswas used. Short echo time examina-tion lasted 3.12 mins, the longer TEmeasurements lasted 6.24 mins each.The voxel sizes were 15x15x15 mm3

and 18x16x20 mm3 respectively. MRspectroscopy was also performed inthe presumably normal left occipitallobe for comparison purposes. Inmeasurements with echo-times of 30 ms and 135 ms, the integral valueof NAA and creatine peaks werelower than normal. All spectra showedan increase in choline. At 1.3 ppm,which represents lactate/lipid, inver-sion of the peak was seen at TE of135 ms. At 30 ms, lipid peak hadsignificantly increased. At 270 ms, no significant peak at this locationwas seen. The NAA/Cho ratio at 135 ms echo-time was lower than fornormal brain tissue. Additionaldiffusion weighted sequences with b = 1000 s/mm2 showed the lesion tobe hypo-intense which was also an indication that it could be tumorrather than abscess.

Until recently, differentiating bet-ween inflammatory and neoplasticmass lesions has been difficult.Nowadays, however, the comple-mentary information received fromMR spectroscopy and diffusionweighted imaging has helped toimprove and even alter the diagnosisof an intracranial mass lesion.

Results and Discussion

Further examinations revealed masslesions in the breast with multiplelesions in the lung, which could bemetastasis with primary lesion in thebreast. Before resection of the lesionwith neuro-navigation, functionalimaging with a simple optical para-digm was also performed. The pa-tient was asked to close and open hereyes 3 times. Each phase – rest andactivation – was scanned 10 timeswith an EPI sequence, 36 slices, with aslice thickness of 3 mm. The resultingBOLD images were fused with T1weighted images. These images werealso used in the neuro-navigationsystem applied during the operation.

The histological analysis of theresected lesion indicated metastasisof papillary cancer such as breast orovarian cancer.

The diagnosis of brain abscessesis generally based on clinical findingsand image findings from CT and MR. Great care is required with imagefindings due to the fact that abscessesand cystic-necrotic tumors can lookvery similar and differential diagnosisis not always easy. As in this case, MRspectroscopy and diffusion weightedimages can bring complementaryinformation. Cystic tumors andabscesses show increased lactatepeak. Lactate is a non-specific meta-bolite that results from anaerobicglycolysis. The spectra from cystic,necrotic brain tumors differ fromabscesses. Abscesses show an increaseof acetate and succinate, amino acidsand lactate due to increased glyco-lysis. Amino acids like valine andleucine are end products from enzymesthat are released from neutrophils in the abscess. It is very important todifferentiate amino acids (valine,leucine and isoleucine at 0.9 ppm)and lipids (0.8-1.2 ppm). Lipids canbe seen in tumors and abscesseswhereas, on the other hand, in vivo


CLINICALMRS BRAIN

Figure 1 Pathologic enhancementof the left facial nerve due to inflam-matory changes in the petrous bone.1b shows normal appearance of theright facial nerve. Oblique coronalviews from 3D MP-RAGE data set.

Figure 2 Mass lesion in the occipital lobe with peripheral enhancement and central liquid part. Differential diagnosis included abscess or tumor. (a) T2 weighted TurboSE sequence; (b) T1 weighted spin echo sequence before contrast; (c) transverse reconstruction from gadolinium-enhanced 3D MP-RAGE data set.

Figure 2a Figure 2b Figure 2c

Figure 1a Figure 1b


CLINICALMRS BRAIN

Figure 3 Single voxel MR spectroscopy. Shown are three measurements with echo times of135 ms, 30 ms. B shows the spectrum from the normal left side. In the metastatic lesion onecan see decreased NAA and increased choline, as well as increased lipid peaks.

a

b

c


CLINICALMRS BRAIN

Figure 4 Preoperative fMRI with an easy paradigm (eye closing andopening). The images show activa-tion in the visual areas in bothoccipital lobes, which will helpduring the operation in sparing theseimportant anatomical landmarks.

Figure 5 Diffusion weighted imageshows central part of the lesion ashypointense, which is an indicator ofmetastatic or tumor lesion.

analysis of tumors will not show anyamino acids except in in vitro condi-tions. At an echo time of 135 ms, the lactate doublet and the aminoacid multiplets are inverted due tothe j-coupling effect. This inversion is not to be found with lipids. This isan additional helpful finding along with the lack of acetate and succinatein tumors. Care must be taken thatthese findings of an abscess are accu-rate before giving antibiotic therapyas there seems to be a decrease inacetate and succinate followingantibiotic therapy. For differentiationof cystic-necrotic tumors and brainabscesses, diffusion weighted imagesare also very useful. The pus in brainabscesses is hyperintense in contrastto the cystic or necrotic parts oftumors. This finding is not 100%reliable, however, as there have beensome reports in the literature showinghyperintense tumor lesions. In theoccipital lesion of the patient therewas an increased lipid peak, whichindicated a metastatic lesion. (Inmetastatic lesions, a higher contrastof lipid may be found in comparisonto glioblastomas, anaplastic astrocy-tomas, abscesses or other intracranialmass lesions.)

The use of MR spectroscopy anddiffusion weighted imaging willimprove the diagnostic certainty ofimaging techniques, which in turnwill decrease the number of interven-tional approaches such as surgeryand biopsy.

the inherent heterogeneity of theselesions which can be useful for biopsyguidance and for tumor extensionpast obvious gross anatomic bounda-ries on MRI.

In the case being presented,conventional MR images (Figs. 1aand b) demonstrate a solid and cysticmass with heterogeneous enhance-ment and mild peritumoral edema.Metabolite maps obtained with longTE CSI (Figs. 2a, 2b, 2c) show a veryhigh Cho:NAA ratio consistent withviable solid tumor, low creatine (Cr)and high lactate. The metaboliteabnormalities closely correlate to theanatomic boundaries of the mass.The single voxel short TE MRS (Figs.3a and b) demonstrates moderatemyoinositol (mI) elevation and very high lipids. This combination ofmetabolite patterns on short andlong TE MRS has been observed intumors of high grade. MRS offers us a level of specificity that could not be achieved with conventional MRIalone.


CLINICALMRS BRAIN

Richard Jones, Ph.D.Susan Palasis, M.D.

Dept. of Radiology,Children’s Healthcare of Atlanta, Atlanta, GA, USA

Proton Magnetic Resonance Spectroscopy(MRS) in Primary Pediatric Brain Tumors

History

This is a seven-year-old male with arecent history of headaches. Thechild developed a new onset seizureand was sent for neuroimagingevaluation. A non-enhanced CT scanof the head demonstrated a masslesion in the left medial occipitallobe. A plain and contrast enhancedMRI of the brain was performedwhich confirmed the presence of a brain tumor. MR spectroscopy was performed for further tumor characterization.

Technique

The data was acquired on a 1.5 TeslaMAGNETOM Symphony scannerusing the standard head coil. The CSI was acquired using TR/TE =1700/270 and a 16x16 matrix, withfour averages and elliptical samplingthe total scan time was 8 minutes andthree seconds. Suppression bands onall sides of the volume were used tosuppress signal outside of the regionof interest.

Results and Discussion

The diagnosis of tumor with MRS isoften straightforward with a typicalspectrum of low N-acetyl-aspartate(NAA) and high choline (Cho) obser-ved on long TE evaluation. This is use-ful adjunctive information to conven-tional imaging based diagnosis. A major benefit of spectroscopy, how-ever, is its ability to further analyzetumor physiology. Low grade andhigh grade tumors can have a similarMRI appearance, but often have avery different MRS appearance. MRShas moved beyond the single voxellong TE examination. Chemical shiftimaging (CSI) allows us to look at the entire extent of the tumor andthe surrounding tissues. Metabolitemaps allow us to readily appreciate

Evaluation of intracranial pathologyis optimally performed with MRimaging. Based on conventionalimaging features alone, it is oftenpossible to differentially diagnosebetween intracranial masses. MRspectroscopy has shown itself to be a useful adjunctive tool in furthercharacterization of pathology.Characteristic spectral patterns havethe ability to differentiate betweenintracranial neoplasms and other masslesions. With the advent of moreadvanced MRS techniques, such aschemical shift imaging, we now havethe ability to more completely evaluatethe chemical make-up of the tumor.Combining the information wereceive from both long and short TEproton MRS, tumor grade can beinferred.


CLINICALMRS BRAIN

Figure 1a Figure 1b

Figure 2a Figure 2b

Figure 3a Figure 3b

Figure 2c

Location and Date of Scan: University Medical Center Nijmegen,Nijmegen, The Netherlands,September 2003

Scanner Model used:Siemens MAGNETOM Trio

Coil used: CP head coil

Software Version: NUMARIS 4 syngo MR 2003T (versionVA23A) + spectroscopy option


CLINICALMRS BRAIN

Marinette van der Graaf, Ph.D.MR Spectroscopist

Department of Radiology,University Medical Center Nijmegen, The Netherlands

Case Report: 3TSjögren-Larsson Syndrome

Image Findings(Figs. 1, 2, 3, 4, 5)

MR images showed only mild signs of cerebral atrophy and subtle whitematter involvement with a smallperiventricular lesion with highsignal intensity on a T2-weightedimage in the left parieto-occipitalregion.

The MR spectrum of the graymatter showed a pattern that is nor-mally obtained at 3T using an echotime of 136 ms. However, the MRspectrum of the white matter showedclearly an additional sharp “lipid”signal at 1.3 ppm in addition to aslightly reduced NAA signal.

Discussion

MR imaging hardly showed anyabnormality that could be in line withSjögren-Larsson syndrome. MRspectroscopy of white matter showeda resonance at 1.3 ppm in the whitematter spectrum, which was absentin the gray matter spectrum. Thesefindings are highly characteristic for Sjögren-Larsson syndrome (seeM.A.A.P. Willemsen et al., Am J Neuro-radiol 25, April 2004, in press). Basedupon these results, further molecularstudies were requested: these finallyproved the diagnosis of Sjögren-Larsson syndrome by demonstrationof mutations in the gene involved inthis disorder.

Patient History

A 33-year-old man was known tohave had a clinical diagnosis ofSjögren-Larsson syndrome (an inbornerror of metabolism causing con-genital skin disorder, spasticity, andmental retardation) since infancy.Because of genetic counseling in thefamily, an attempt was made toprove the diagnosis at the molecularlevel. Unexpectedly, repeated bio-chemical investigations were negative.Because of the high clinical suspicion,additional investigations wererequested, including cerebral MRimaging and spectroscopy.


CLINICALMRS BRAIN

Figure 1 T2 weighted image (TE/TR = 104/4000).

Figure 2 Voxel position gray matter.

Figure 4 Voxel position whitematter.

Figure 5 Proton MR spectrum(PRESS, TE/TR = 136/2000) of a 8-mlvoxel located mainly in white matterin the occipital trigone.

Figure 3 Proton MR spectrum(PRESS, TE/TR = 136/2000) of a 8-mlvoxel located in the central occipitalgray matter.

NAA

Cho

Cr

Cr

NAA

“lipid”

Cho

Cr

Cr

axial, coronal) gapless heavily T2-weighted (turbo factor 15) turbo spinecho (TSE) data sets within 2 minutesfor 3-dimensional positioning of thevolume of interest (VOI).

Since it has been shown thatoccipital gray and white and parietalwhite matter locations exhibit essen-tially identical MRS metabolite chan-ges in HE [1], we decided to place theVOI in the medial part of the occipitallobe. This area contains grey andwhite matter and allows the measu-rement of a VOI size of 2x2x2 cm3

with good homogeneity of themagnetic field (Fig. 1).

MR imaging consisted of non-enhanced axial T1 weighted (T1w)spin echo (SE) and T2 weighted(T2w) TSE images. Measurementparameters (TR/TE/acquisition time)were 580ms / 12ms / 2min for theT1w and 3000ms / 90ms / 3min forthe T2w imaging. Slice thickness SLwas 5 mm with a gap of 1 mm.

In chronic liver disease, signalhyperintensities in the basal gangliaon T1w images are found mostpronounced in the globus pallidus[4]. They probably represent manga-nese deposition due to liver dysfunc-tion which is potentially reversible ifliver function improves or wherethere has been a liver transplant [5].

For the detection and quantifica-tion of mI and Glx, it is essential touse spectroscopic sequences with ashort echo time (TE) to minimize theinfluence of J-coupling and T2 relaxa-tion (Glx, mI), as both factors lead toa signal decrease.


CLINICALMRS BRAIN

Thomas Nägele1, Uwe Seeger1,Holger Hass2, Wilhelm Küker1,Stefan Heckl1, Uwe Klose1

1 Department of Neuroradiology,2 Department of Gastroenterologyand Hepatology Medical Clinic,University of Tübingen

Localized Proton Spectroscopy in Hepatic Encephalopathy: Advantage of 3T-High-Field for Discrimination of Glutamine and Glutamate

panying HE. For comparison, measu-rements were acquired at 3T and1.5T (Siemens Sonata) and thespectra were compared with those ofa healthy volunteer at 3T and 1.5T.

Patient History and Methods

We examined a 56-year-old man withethyltoxic liver cirrhosis (Child-PughC). The referring hepatologist perfor-med a clinical examination for detec-tion of the HE including a numberconnection test (NCT) [2] immediatelybefore the MR examinations. Patho-logical NCT was found with 90 swhile the NCT score can be classifiedas normal provided the time forfinishing the test is less than tentimes the subjects age in decades [3].Additionally, the patient sufferedfrom drowsiness as well as elevatedammonia blood levels during the lastdays in hospital. This made a reliableclinical diagnosis of Hepatic encepha-lopathy possible.

Two subsequent MRS and MRIexaminations of the same patientwere performed within one day on a1.5T whole-body MR System (MAGNETOM Sonata, Siemens AG,Erlangen, Germany) using a conven-tional circularly polarized head coiland on a 3T whole-body MR System.(MAGNETOM Trio, Siemens AG,Erlangen, Germany) using a standardhead coil. For the follow-up examina-tion, exact reproducibility of thespectral localization was of specialimportance. This was achieved byacquiring 3 orthogonal (sagittal,

IntroductionIn hepatic encephalopathy (HE) – a frequent complication of chronicliver dysfunction – the neuro-psycho-logical impairments range from milddeficits in psychomotor and visio-practic abilities to confusion andfinally stupor in higher grades due todecreased hepatocellular detoxifi-cation and synthesis functions. In HE,specific changes of brain tissueconcentrations of myo-Inositol (mI)(decrease), glutamine/glutamate(Glx) (increase) and Choline (Cho)(decrease) have been reported [1].For spectroscopic detection of Glx,short echo times and highest possiblefield strength are preferred. In thisstudy, therefore, we used a shortecho time PRESS sequence (TE = 30 ms)to achieve high signal of the J-coup-ling resonances mI and Glx. Ofprincipal interest is the differentiationof glutamine (Gln) and glutamate(Glu) which is generally very difficultto achieve at the standard fieldstrength of 1.5T. This report showsthat this differentiation is possible ata whole body unit of 3T (SiemensTrio), as demonstrated in a patientwith chronic liver disease and accom-


CLINICALMRS BRAIN

Figure 1 T1w MR image acquired at1.5T showing the voxel localization(white box) in the occipital gray/whi-te matter. Further signal hyperinten-sities of the basal ganglia (globuspallidus) are depicted (white arrow)representing manganese depositiondue to liver dysfunction which ispotentiality reversible if liver func-tion improves or where there hasbeen a liver transplant.

Figure 2 Gray matter short echo time spectra from a healthy volunteerobtained at 1.5T and 3T after evaluation by LCModel. Additionally, theproportions of glutamate and glutamine obtained are shown by scaled modelspectra.

3 Tesla

mlCho

Cr

NAA

ppm ppm

glutamate glutamate

glutamine glutamine

ml Cho

Cr

NAA

1.5 Tesla

healthy volunteer

4 3 2 1 4 3 2 1 0

bolisms of the excitatory neurotrans-mitter glutamate should be possible.This might be of special interest fordetection of the mild and subclinicalforms of HE which are characterizedby missing clinical symptoms. Never-theless, these patients are unable to drive a vehicle due to reducedvisiopractic abilities [1]. Up to now,subclinical HE is commonly diagno-sed by extensive neuropsychologicaltesting. However, all these tests aremore or less dependent on theinvestigator and on the patient’s age,sex and education.

References [ 1 ] Ross B D, Jacobson S, Villamil F, Korula J,Kreis R, Ernst T, et al. Subclinical hepatic encephalopathy: proton MR spectroscopic abnormalities.Radiology 1994;193:457-463.

[ 2 ] Parsons-Smith B, Summerskill W, Dawson A, Sherlock S. The electroencephalograph in liver disease.Lancet 1957;2:867-871.

[ 3 ] Van Gorp W, Satz P, Mitrushina M.Neuropsychological processes associated withnormal aging. Dev Neurpsychol 1990;6:279-290.

[ 4 ] Brunberg JA, Kanal E, Hirsch W, Van Thiel DH. Chronic acquired hepatic failure: MR imaging of the brain at 1.5. AJNR Am J Neuroradiol 1991;12:909-914.

[ 5 ] Naegele T, Grodd W, Viebahn R, Seeger U,Klose U, Seitz D, Kaiser S, Mader I, Mayer J,Lauchart W, Gregor M, Voigt K: MR imaging and 1H spectroscopy of brainmetabolites in hepatic encephalopathy: time-course of renormalization after livertransplantation. Radiology 2000; 216:683 – 691

[ 6 ] Provencher S W: Estimation of metabolite concentrations fromlocalized in vivo proton NMR spectra. Magn-Reson-Med. 1993; 30: 672-679

[ 7 ] Seeger U, Klose U, Mader I, Grodd W,Nägele T: Parameterized evaluation of macromoleculesand lipids in proton MR spectroscopy of braindiseases. Magn Reson Med 2003; 49: 19-28


CLINICALMRS BRAIN

glutamine concentration [5.3 mMol/l]in the patient leads to a similar signalstrength of Glu and Gln. Togetherwith the higher chemical-shift at 3T,proper separation and reliable quanti-fication of glutamine and glutamateis possible with the LC-model eva-luation (Fig. 4). Usually this is notpossible at 1.5T, leading to frequentlyused combined evaluation of bothmetabo-lites as a resonanceGlx=Glu+Gln at lower field-strengths.

Discussion

This report shows an example ofpractical benefit of the theoreticallypredicted higher chemical shift at 3Tcompared to 1.5T. Separation ofglutamine and glutamate is of majorimportance for the understanding ofHE, which is an important complica-tion of chronic liver dysfunction. 70% of patients with chronic liverdysfunction suffer from HE episodi-cally. The symptoms range from milddeficits in psychomotor and visio-practic abilities to confusion andfinally stupor in higher grades due todecreased hepatocellular detoxifica-tion and synthesis functions.

High ammonia levels due toreduced detoxification of the liver isthought to be a reason for elevatedcerebral glutamine concentrationsfound in previous spectroscopicstudies [5]. These studies were eitherbased on plain peak integrationbetween 3.72 ppm and 3.82 ppm toevaluate a-Glx [1] – as it has beenfound that the integrated signalcorresponds well to the predominantglutamine signal – or on fittingprocedures with subsequent peak integral calculations [5].

In principle, none of these vivostudies at 1.5T allow separation ofglutamine and glutamate. However,if glutamine and glutamate can beseparately quantified, then furtherinteresting insights into the meta-

Therefore 1H-MRS was performedusing a single voxel double spin echosequence (PRESS, TR 3000, TE 30, NA = 64) with a measurement timeof 3minutes per voxel. For each voxel a short reference measurement with-out water suppression was performedfor an eddy current correction duringpostprocessing.

Spectral postprocessing wasperformed using an optimized LC-model [6] for absolute quantification.In our LC-model, the basis data setwas completed by lipid and macro-molecule resonances which areespecially important in demyelinatingand neoplastic lesions [7]. The resultsof the LC-model evaluation aredepicted in Figs. 2 and 3 for bothmeasurements at 1.5 Tesla and 3 Tesla. The left/right column in Fig. 2shows the results for 3T/1.5T of thehealthy volunteer. The upper rowdepicts the LC-model fit of the spectra,the middle row the glutamate (Glu)and the lower row the glutamine(Gln) signal. It is clearly shown that innormals cerebral concentration ofGln is relatively low [1.9 mMol/l] incomparison to the excitatory neuro-transmitter (Glu) [6.5 mMol/l].

Of major importance is the sepa-ration of the �- (3.77 ppm) and �-(2.05-2.45 ppm) resonances of Glnand Glu at 3T due to increased chemical-shift while they are slightly togetherat 1.5T. This increased chemical-shiftallows better separation of Glu andGln in the LC-model fit at 3T comparedto 1.5T.

More prominently, this is depictedin Fig. 3 with the correspondingspectra of the patient with hepaticencephalopathy. The upper rowshows again the fit of the LC-modelwith an HE-typical decrease of choline(Cho) and myo-inositol (mI). Additio-nally, in comparison with Fig. 2 anelevated glutamine signal can bedetected, best seen in the range ofthe b- and g-resonances. The elevated


CLINICALMRS BRAIN

Figure 3 Gray matter short echo time spectra from a patient with HW obtained at 1,5T and 3T after evaluation by LCModel. Additionally, the proportions of glutamate and glutamine obtained are shown by scaled modelspectra.

Figure 4 The certainty of the quantification of Gln and Glu is higher at 3Tcompared to 1.5T. If these two metabolites have to be determined separately,higher field strengths are advantageous, whereas a combined evaluation asGlx is also possible at 1.5T. Especially for Gln, the 20% level that indicates lowcertitude of the determination in the LC-model is almost reached at 1.5T evenin the case of HE when Gln is elevated, which should simplify its determination.

3 Tesla

stan

dard

dev

iati

on

Glu + Gln = Glx Gln Glu

Sonata 1.5T Trio 3.0T

ml

Cho

Cr

NAA

ppm ppm

glutamate glutamate

glutamine glutamine

ml

Cho

Cr

NAA

1.5 Tesla

patient with HE

comparison of 1.5Tesla and 3Tesla

4

25%

20%

15%

10%

5%

0%

3 2 1 4 3 2 1 0

related to both cancer as well as non-malignant abnormalities such asBPH, prostatitis, biopsy-induced intra-parenchymal hemorrhage or therapy-induced tissue degeneration.

Beyond morphologic informationprovided by MR imaging, in vivoproton MR spectroscopy enables tononinvasively detect small mobilebiomolecules, which are fingerprintsof prostatic metabolism. The wholeprostate can be covered with voxelvolumes of less than 0.5 ml using 3DMR spectroscopic imaging (MRSI).Combined MRI/MRSI provides high-resolution anatomic and metabolicinformation enabling better tumorvisualization within the prostate.Within each voxel the resonances ofcitrate (Cit), choline-containingcompounds (Cho), and the creatine-phosphocrea-tine complex (Cr) canbe detected and relative signalintensity ratios can easily be deter-mined using software provided bythe manufacturer. In prostate cancerincreased signal intensities of Chocompared to Cit can be observed.Based on this characteristic metabolicfeature of cancer tissue, clinicalstudies have shown that combinedMRI/MRSI enables improved tumordetection and characterization [1]. Inparticular, increased diagnosticaccuracy and decreased inter-obser-ver variability have been found forthe diagnosis of extracapsular exten-sion [3] [4]. Moreover, the combineduse of MRI/ MRSI increases the overallaccuracy of determining prostatecancer volume, which is a provenprognostic factor [5]. Metabolicinformation of MRSI may also indicateindividual tumor aggressiveness,since the increase of Cho and thedecrease of Cit signal intensities havebeen found to correlated with theGleason score [1].


CLINICALMRS PROSTATE

Heinz-Peter Schlemmer, M.D., Ph.D.

University Hospital Tübingen,Dept. of Radiology, Tübingen,Germany

Proton MR Spectroscopic Imaging in theClinical Evaluation of Prostate Cancer

The preferred treatment of earlystage cancer confined to the prostatestage is radical prostatectomy. Anaccurate preoperative identificationof extracapsular tumor extension,seminal vesicle invasion and distanttumor spread is important to selectsurgical candidates and to avertunnecessary surgery. It is importantto note, however, that side effects ofthe treatment adversely affect qualityof life, particularly in previouslyasymptomatic patients with smalltumors confined to the gland. Nume-rous alternative therapeutic approa-ches have therefore been investigated,which are less invasive and havereduced risk of developing complica-tions such as impotence and inconti-nence compared to surgery (e.g.stereotactic radiotherapy, androgendeprivation, ‘watchful waiting’). Datafrom randomised clinical trials com-paring the cancer control rates arenot yet available, however. For indi-vidual treatment, planing monitoringand improved methods are necessaryto characterize individual tumoraggressiveness particularly in prostatecancer, which is characterized by a high biological variability. Only asmall proportion of men with untreatedcarcinoma will develop serious morbi-dity or will die from the disease.

For local tumor staging andcharacterization, the accuracy ofmost commonly used clinical para-meters (digital rectal examination, PSAserum level and Gleason grading) islimited. The role of clinical imagingfor local staging and tumor charac-terization is dramatically evolving.MR imaging with combined endorec-tal and phased array coil is mostchallenging for assessing extrapro-static tumour extension and seminalvesicle infiltration with reportedaccuracies of up to about 80% and95%, respectively [1] [2]. The accuracyis limited, however, because abnormalsignal within the prostate can be

Introduction

Prostate cancer has become a leadingcause of morbidity and mortality inmen in Western countries. Due to theincreasing natural life expectancy, afurther increase of the incidence canbe expected over the next decades.

Early diagnosis is significantlyimproved using PSA serum testing.PSA levels have also been shown tocorrelate with intraprostatic andextraprostatic tumor volume, whichare of prognostic importance. Theaccuracy of mid-range PSA levels ofabout 4.0 to 10.0 ng/ml in a singlepatient is limited, however, becausemuch overlap exists in men withprostatic cancer and benign prostatichyperplasia (BPH). Accordingly,systematic needle biopsy has to beperformed in any case to finallyconfirm the presence of cancer andto assess the tumor grade.



Diagnostic MRSI as an adjunct to conventional MRI can be indicatedin four main categories:1) Evaluation of patients with

suspicious PSA-levels but negativeprior sextant biopsies: improvedtumor localization and support fortargeted biopsy [2].

2) Pre-operative evaluation of patients with biopsy-provenprostate cancer: assessment oftumor volume and extraprostatictumor spread [2] for individualtherapy planning (e.g. radio-therapy).

3) Monitoring effects of minimalinvasive treatment strategies.

4) Evaluation of patients with risingPSA-levels during follow-up aftertherapy: Improved detection andlocalization of local recurrence.

Technical Prerequisites

For successfully monitoring themetabolic state of the prostate, thefollowing prerequisites must befulfilled:1) Availablity of 3D CSI for covering

the entire organ.2) A good B0-homogeneity of the

area of interest must be reachedpreferably by an automized shimming procedure. This is the prerequisite for proper function of other technical features such as water and lipid suppression,and signal separation during post-processing.

3) Sufficient signal to noise.4) Endorectal Coil.

The metabolic signals of interestsare typically 104 times weaker thanthe water signal used for MRI, due tothe lower molecular concentrations.Hence it is of paramount importanceto collect sufficient signal. The totalscan time is usually given by clinicalfeasibility. Optimal usage of the scantime is achieved by using weightedacquisition, i.e. spending more timeon collecting central than peripheralk-space data; e.g. while a 8x8x8acquisition of a TR = 1.5 s and fouraverages requires more than 50 minutes using conventional full k-space sampling, the measurementtime for acquiring data of the sameSNR is reduced to approximately 8 minutes using weighted acquisition.The spatial resolution of weightedacquisition is determined by a filter-function which increases the widthof the central lobe of the spatialresponse function (“effective voxelsize”), but which reduces the signalcontributions from distant voxels(“voxel bleeding”). For obtaining thelatter, this filter function is oftenapplied even to fully sampled k-spacedata (which leads, however, to asevere loss of SNR). For many clinicalquestions it is important not tosacrifice further the already low spatialresolution of MRS for increasing SNR.Instead, there is a tendency of usingan endorectal coil for receiving closeto the prostate the highest signals.Using this setup, effective voxel sizesof approximately 1cc can be reachedfor an acquisition time of 10 minuteson a 1.5T scanner.

Future considerations

An ongoing clinical study initiated by Siemens (International MulticenterAssessment of Prostate Spectroscopy,IMAPS) has the goal of proving thevalue of the described MRS techni-ques for tumor visualization andlocalization. Furthermore, larger-

References [ 1 ] Kurhanewicz J, Swanson MG, Nelson SJ,Vigneron DB. Combined magnetic resonanceimaging and spectroscopic imaging approachto molecular imaging of prostate cancer. J Magn Reson Imaging 2002; 16(4): 451-63.

[ 2 ] Engelbrecht MR, Jager GJ, Laheij RJ,Verbeek ALM, van Lier HJ, Barentsz JO. Local staging of prostate cancer using magneticresonance imaging: a meta analysis. Eur Radiol 2002; 12: 2294-2302.

[ 3 ] Yu KK, Scheidler J, Hricak H, Vigneron DB, Zaloudek CJ, Males RG, Nelson SJ,Carroll PR, Kurhanewicz J. Prostate cancer: prediction of extracapsularextension with endorectal MR imaging and three-dimensional proton MR spectroscopicimaging. Radiology 1999; 213(2): 481-488.

[ 4 ] Scheidler J, Hricak H, Vigneron DB, Yu KK,Sokolov DL, Huang LR, Zaloudek CJ, Nelson SJ,Carroll PR, Kurhanewicz J. Prostate cancer: localization with three-dimensional proton MR spectroscopic imaging-clinicopathologic study. Radiology 1999; 213(2): 473-480.

[ 5 ] Coakley FV, Kurhanewicz J, Lu Ying, Jones KD, Swanson MG, Chang SD, Carrol PR,Hricak H. Prostate cancer tumor volume:measurements with endorectal MR and MRspectroscopic imaging. Radiology 2002; 223; 91-97.

scale clinical trials have to be perfor-med to determine the clinical valueof combined MRI/MRSI studies for themanagement of prostate cancerpatients.

Increased spatial and spectralresolution can be expected by usingMR scanners with 3 Tesla or evenhigher magnetic fields. More anato-mic details and probably new meta-bolic markers may further increasethe diagnostic accuracy and probablythe assessment of individual aggres-siveness.



Results from IMAPS Study

Image Findings (Results)

The localization of the tumor-regionin the histopathologic examinationof the resected prostate (Fig. 1) isreflected in the MRS data (Fig. 2).Tumor spectra are characterized byhigh (cholin+creatin)/citrate ratios(Fig. 2b).

Measurement Details

• 3D_CSI_SE_prostate_specialsequence with spectral lipid andwater saturation (MEGA-pulses)*

• endorectal coil• automatic adjustments• acquisition time = 10:41 min• TR = 650 ms, TE = 120 ms,

a = 900, vectorsize = 512 samples,acquisition bandwidth = 1250 Hz,nominal voxel size = 4 x 5 x 5 mm,matrix size = 16 x 16 x 16, weighted acquisition

• Sonata, software MR 2002B

Discussion

This case is an example from theIMAPS study (International Multi-Centre Assessment of Prostate MRSpectroscopy) which investigates the capabilities of 1H CSI to detectprostate carcinoma in vivo. For thiscase the MRS-examination confirmsthe existence and spread of thecancerous tissue. Regarding spectralquality this is a borderline case: data quality is just sufficient for thecase to be included. Data qualitycould be improved by adjusting the

Figure 1 Histopathology resultindicated prostate carcinoma. The cancerous region is marked.

Figure 2 MRS results, which arefrom the same prostate section asthe histopathology slice of Fig. 1.Figure 2a Axial reference image.

Figure 2b Metabolite map of theratio (choline+creatine)/citrate. Its expansion is indicated by the blue rectangle in a) Tumor spectraare characterized by high (cholin+creatin)/citrate ratios.

Figure 2c Spectral map, whichreflects the alteration of the meta-bolite composition from suspicioustissue (lower left part with highercholine level) to inconspicuousfindings.

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

* WIP: The information about this product ispreliminary. The product is under developmentand is not commercially available in the US andits future availability cannot be ensured.

saturation slabs more accurately and thereby eliminating lipid conta-mination.

Courtesy of Dr. Matthias Lichy Universitätskrankenhaus Tübingen, Radiologie



Image Findings (Results)

The localization of the tumor-regionin the histopathologic examinationof the resected prostate (Fig. 1) isreflected in the MRS data (Fig. 2).Tumor spectra are characterized byhigh (cholin+creatin)/citrate ratios(Fig.2b).

Measurement Details

• 3D_CSI_SE_prostate_special sequence with spectral lipid andwater saturation (MEGA-pulses)*

• endorectal coil• acquisition time = 10:27 min• TR = 650 ms, TE = 120 ms, a = 900,

vectorsize = 512 samples, acquisition bandwidth = 1250 Hz,nominal voxel size = 5 x 5 x 4 mm,matrix size = 16 x 16 x 16, weighted acquisition,

• Sonata, software MR 2002B

Discussion

This case is an example from theIMAPS study (International Multi-Centre Assessment of Prostate MRSpectroscopy) which investigates the capabilities of 1H CSI to detectprostate carcinoma in vivo. For thiscase the MRS-examination confirmsthe existence and spread of thecancerous tissue. As the MRS data isabsolutely free from mayor lipidcontamination and of excellentspectral resolution, it is an outstandingdemonstration of the data qualitywhich is achievable with this technique.

Figure 1 Histopathology resultindicated adenocarcinoma of theprostate. The cancerous region ismarked.

Figure 2 MRS results from the sameprostate section as the histopathologicslice of Fig. 1.Figure 2a T2 weighted axial referenceimage, which reveals a hypo intensespot in the region of the lesion.

Figure 2b Metabolite map of theratio (choline+creatine)/citrate.Tumor spectra are characterized byhigh (cholin+creatin)/citrate ratios.

Figure 2c Spectral map, which reflectsthe alteration of the metabolite com-position from suspicious tissue (centralright part with higher choline levelsand lower citrate levels) to inconspi-cuous tissue. Its expansion is indicatedby the white rectangle in b). (Cit: Citrate, Cho: Choline, Cr: Creatine).

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

* WIP: The information about this product ispreliminary. The product is under developmentand is not commercially available in the US andits future availability cannot be ensured.

Courtesy of Dr. Matthias Lichy, Universitätskrankenhaus Tübingen Radiologie

Figure 1a Spectral map (range 1.5to 4 ppm) from one out of 16 slicesfrom the 3D MRSI dataset, overlaidon an axial T2-weighted image ofthe prostate of the patient. Thevoxels indicated in blue are enlargedin Fig. 1b and Fig. 1c.

Figure 1b Spectrum from tissue inthe healthy peripheral zone of theprostate. The citrate signal is cente-red at 2.60 ppm, the choline signalat 3.20 ppm overlaps with the creatinesignal at 3.04 ppm.

Figure 1c Spectrum from tumourtissue. Notice the relative increase incholine, and decrease in citratesignals.



Dr. Tom Scheenen

University Medical Centre NijmegenDept. of RadiologyNijmegen, The Netherlands

Case Report: Prostate Carcinoma Stage T3b

operative PSA was 0.1 and the patienthad recovered erection. The possibi-lity of recurrence is monitored.

Measurement Details

• 3D_CSI_SE_prostate_special sequence with spectral lipid andwater saturation (MEGA-pulses)

• endorectal coil• automatic adjustments, of which

shimming was repeated twice• acquisition time = 12:40 min• TR = 650 ms, TE = 120 ms, a = 900,

vectorsize = 512 samples, acquisition bandwidth = 1250 Hz, acquisition-weighted k-spacesampling with 6 averages in thecenter of k-space, field of view 70 x 60 x 60 mm, matrix size = 14 x 12 x 12, filtered and zerofilledto 16 x 16 x 16 matrix.

Sonata, software MR 2002B

Discussion

This case is an example from theIMAPS study (International Multi-Centre Assessment of Prostate MRSpectroscopy) which investigates thecapabilities of 1H CSI to detect prostate carcinoma in vivo. Despitethe small chance for seminal invasionbased on the Partin tables, the combination of MRI and MRSI predic-ted extension of the tumour into theseminal vesicles, which was confirmedby histopathology of the removedprostate. In this case the MRS-exami-nation confirms the existence andspread of the cancerous tissue,whereas also healthy tissue can becharacterized as such.

Patient History

Age: 62 years, PSA = 5.9, Biopsy:Gleason 6/10, Suspected stage: T1cWith his stage, PSA and Gleasonscore the pretest probability ofhaving limited disease (confined tothe prostate) is 71%, according to thePartin coefficient tables. Upon perso-nal request, we called this patient,who was under treatment in a diffe-rent institution, for an MRI/MRSIexam.

Image Findings

We found a large tumor predominantlyin the left peripheral zone, minimallyextending into the left seminalvesicle (T3b). The tumor showsrelatively high choline over citrateratios (which is a pathologic marker),also centrally around the urethra.There is a smaller limited tumor moreclose to the apex without extracap-sular spread.The tumors show somepathological enhancement. Basedupon MRI/MRS findings, the diagno-sis is stage T3b.

At the patient’s treating institution,the urologist performed a prostateand seminal vesicles resection withintrafacial nerve sparing on the rightside and an extra facial nerve sparingon the left side. The pathology of thispatient was indeed pT3b. The patho-logists found an extra prostatic exten-sion in front of the left and rightbasis, and the left seminal extensionthat was detected with MRI/MRS. The margins were negative. The post



Figure 2a Axial T2-weighted imagethrough the base of the prostate. The white box is the volume ofinterest, or the PRESS box, of the 3DMRSI measurement.

Figure 3a to c Three metaboliteratio images of the base of theprostate, overlaid on adjacent T2-weighted images. Deviating Cholineover citrate ratios are detected in the center of the prostate, extendingto the left peripheral zone towardsthe seminal vesicles.

Figure 2b Coronal T2-weightedimage of the prostate and surroun-ding tissues. A hypo-intense lesionextends from the prostate into theleft seminal vesicles.

a

b

c

Figure 1a Voxel location of spectrum of healthy tissue.

Figure 1b Spectrum of healthy tissue.



O. Soellner M.D.1, W. Pegios M.D.1, Th.J. Vogl M.D.1, M. Wolfram M.D.2, D. Jonas M.D.2

1 Dept. Diagn. and Intervent. Radiology, 2 Dept. UrologyUniversity Hospital Frankfurt a.M.,Germany

Prostate Carcinoma Detected with Single Voxel Spectroscopy

The patient had a PSA-value of 13.2.The MR exam was carried out after a first biopsy of the prostate showinga negative result. While the left handside of the prostate appeared normalin the spectrum (Figs. 1a and b), the spectrum of the right hand sideshowed clearly elevated choline andreduced citrate levels (Figs. 2a and b).The spectral resolution achieved by automized shimming is excellent.A second biopsy confirmed a trabe-cular prostate adeno-carcinomabetween G2 to G3 (combined Gleason-grade 10).

This is one of our early prostateMRS examinations, carried out at atime when the prostate CSI packagewas not yet available. While 3D-CSI is the method-of-choice used today,this case still shows that even an SVSexam can lead to clinically relevantresults.



Figure 2a Voxel location of spectrum of suspected carcinoma.

Figure 2b Spectrum of suspected prostate carcinoma.Elevated choline, reduced citrate.

Figure 1a Axial T2 weighted TSEimage (TE 132 ms) of the prostate ofa patient with prostate cancer at 1.5 T. The white box indicates thePRESS box selection of the 3D MRSImeasurement; voxels I and II are thepositions of figures 1b and 1c.

Figure 1b Spectrum of a voxel inhealthy tissue of the peripheral zoneof the prostate. In the spectral rangeof 1.8 to 3.5 ppm resonances fromcitrate, creatine and choline arepresent. Polyamines could be presentbetween choline and creatine,connecting these resonances witheach other.



Tom W. J. Scheenen, Ph.D. Jurgen J. Fütterer, M.D.Jelle O. Barentsz, M.D., Ph.D.Arend Heerschap, Ph.D.

Department of Radiology,University Medical Center Nijmegen,The Netherlands

1H-MR Spectroscopic Imaging of the Human Prostate: from 1.5 to 3T

The clinical value of prostatespectroscopy at 1.5T with a dedicatedpulse sequence on different MaestroClass scanners is currently beingevaluated. Siemens Medical Solutions,together with Medrad Inc. andaround ten other participating clinicalinstitutions, have organized an Inter-national Multi-centre Assessment ofProstate Spectroscopy (IMAPS) trial,the primary objectives of which areto prove that 3D MRSI of the prostatecan detect and localize prostatecancer on the basis of the ratio of thecholine signal integral over the citratesignal integral (for more information,please look at the IMAPS home pagehttp://get.to/IMAPS).

Since the MR signal of citrate at1.5 and 3T originates from a stronglycoupled spin system, its spectralshape depends on magnetic fieldstrength and PRESS (Point RESolvedSpectroscopy) pulse sequence timing.At an echo time of 120 ms at 1.5T,the citrate resonances appear as a singlet; almost all intensity in twolargely overlapping peaks around2.60 ppm and hardly any intensity inthe two surrounding satellite peaks.However at 3T, the spectral resolutionincreases twofold, more clearlyrevealing the complex shape of thecitrate signals that can appear witheither positive or negative inner linesand non-zero satellite peaks. Afterunderstanding the spectral shape ofcitrate at 3T, the advantages ofmoving prostate spectroscopy to 3Tcan be appreciated. Apart from anincrease in spectral resolution, whichincreases the separation of individualresonances, the SNR of the metaboli-tes also increases, which can be usedto either decrease total measurementtime or increase spatial resolution.

Proton MR spectroscopic imaging (SI) of the prostate is a challenge: theorgan is small, resides in the centreof the male body, is embedded inlipid tissue, and is close to intestinescontaining air. The use of an endo-rectal surface coil, positioned veryclose to the prostate, enables anadequate SNR and spatial resolutionof the MRSI measurement. Lipidsignals are suppressed with a combi-nation of saturation slabs and dualfrequency-selective rf pulses in thepulse sequence. Automated shim-ming algorithms produce acceptableline widths of the water resonancebefore the actual MRSI measurementstarts.

choline

creatine

citrate

Voxel I,Healthy



Figure 1c Spectrum of a voxel inpossible tumor tissue in the centralgland of the prostate. Notice therelative difference in choline andcitrate signal intensities: increasedcholine and decreased citrate, combined in the choline/citrate ratio,are markers for tumor tissue.

Figure 2 Useful theoretical shapesof the citrate resonance at 3T. The three different spectral shapescorrespond to three different PRESSpulse sequence timings: TE of 75,100 and 145 ms respectively.

Figure 3a T2 weighted TSE image(TE 109 ms) of a patient with prostatecancer at 3T. MRSI voxels from oneout of 16 MRSI slabs from the 3Dmatrix are overlaid in green. Thepurple box is enlarged in 3b, the bluevoxel is the origin of the spectrum in 3d.

Figure 3b Enlargement from fig. 3a of 12 nominal voxels. For everyvoxel the corresponding spectrum isshown from 2.0 to 3.5 ppm. Thelarge choline signal intensitiescompared to citrate are indicative for the presence of tumor tissue.

Figure 3c Spectrum from the bluevoxel (Fig. 3a) at 1.5T. The line-widths are small enough to see thesplitting of the citrate resonance,even at 1.5T.

Figure 3d Spectrum from the bluevoxel (Fig. 3a) at 3T. Notice the high SNR of this spectrum, and theinversion of the citrate signal at anecho time of 75 ms at 3T.

citrate

TE 120 ms

1.5T

citrate

TE 75 ms

3T

Voxel II,Tumor

Figure 1 1H in vivo spectrum of alactating breast. tCho refers to totalcholine; MAGNETOM Sonata (1.5T),voxelsize 8 cc, acq. time 3 min 12 sec. Courtesy of Dr. B. Joe and Dr. T. Bae,Washington University School ofMedicine, St. Louis, MO, USA.

Figure 2 tCho signal dedected in ductal carcinoma; MAGNETOM Symphony (1.5T), voxel size 3.4 cc, acq. time 3 min 12 sec. Courtesy of Dr. I. Gribbestad et al., St. Olavs Hospital, Trondheim, Norway.


CLINICALMRS BREAST

Stefan RoellMarianne Vorbuchner

MRS Application DevelopmentSiemens Medical SolutionsErlangen, Germany

The Potential of 1H MRS of the Breast*

signal, and the correction for signalvariations induced by breathing.Currently, a WIP SVS Spin-Echosequence, complemented withadditional spectral (lipid) suppressionpulses, is used for examining thesensitivity of choline detection in the breast.

Today, contrast-enhanced MRI of the breast achieves a high sensitivityfor detecting suspicious masses.However, variable specificities havebeen reported using this technique.MR spectroscopy has the ability todetect metabolic changes in tissueand may provide an added measureof confidence in the characterizationof breast lesions. In vivo investiga-tions of the phospholipid metabolismby 13C- or 31P-MRS appear to berestricted to cases of advancedcancer, due to the low sensitivity ofthese methods. The much highersensitivity of 1H-MRS allows the detec-tion of smaller carcinoma. Using in vivo 1H MRS, the singlet signal ofcholine-containing compounds at 3.2 ppm appears to be significant forthe detection of carcinoma. Other invivo 1H signals of breast tissue originfrom water and lipid compartments.An additional lactose signal is seen in spectra of lactating breast. Whilstit was previously suspected thatcholine is only detectable in malig-nancies, high field studies have nowshown that healthy breast tissue alsoproduces a weak signal at 3.2 ppm.Hence, it is necessary to calibrate thedetected choline signal. One sugges-ted method involves the use of theunsuppressed water signal as aninternal standard, i.e. estimating theconcentration of choline in its solvent(water), thereby correcting for thepartial volume of adipose tissue.Among the other technical require-ments of 1H breast MRS are aneffective suppression of the lipid


lactose

tCho


CLINICALMRS BREAST

principles of 31P-MRS are best explainedin the experimental setting. The mostwidely used animal model for 31P-MRSis the isolated buffer-perfused rodentheart. MRS is performed using a highfield (7-18 Tesla) MR spectrometer.The magnet bore holds the nucleus-specific probe head with the radiofrequency (RF) coils, which are usedfor MR excitation and signal reception.The magnet is interfaced with acontrol computer, a gradient system,and an RF transmitter and receiver.After shimming the magnetic field, aradio frequency impulse is sent intothe RF coils for spin excitation. Theresulting MR signal, the free induc-tion decay (FID) is recorded. Using“Fourier transformation”, the FID isconverted to an MR spectrum, whichrelates resonance frequency andsignal intensity. Due to the lowsensitivity of 31P-MRS, many FIDshave to be signal-averaged to obtainMR spectra with a sufficient signal-to-noise ratio. MR spectra have to becorrected for the effects of partialsaturation.

A typical 31P-MR spectrum from an isolated, beating rat heart isshown in Figure 1 (upper left panel).A 31P-spectrum shows six resonances,corresponding to the three 31P-atomsof ATP, phosphocreatine (PCr),inorganic phosphate (Pi) and mono-phosphate esters (MPE; mostly AMPand glycolytic intermediates). “Chemical shift” (quantified relativeto the B0 field in ppm = parts permillion) describes the phenomenonthat different metabolites resonate atdistinct frequencies, allowing theirdiscrimination from each other. Thearea under each resonance is propor-tional to the amount of each 31P-nucleus in the sample, and metaboliteresonances are therefore quantifiedby integrating peak areas. Relativemetabolite levels are calculateddirectly (such as the phosphocreatine/ATP ratio), but absolute metabolite


CLINICALMRS MULTI NUCLEI / HEART

Stefan Neubauer, M.D. FRCP

University of Oxford Centre for Clinical Magnetic ResonanceResearchUnited Kingdom

31P-MR Spectroscopy of the Heart – Current Status and Future Potential

substrate for all energy-consumingreactions in the cell. Phosphocreatine,the other major high-energy phos-phate compound, acts as an energyreservoir and has at least two addi-tional roles: First, phosphocreatineserves as an energy transport mole-cule in the “creatine kinase / phos-phocreatine energy shuttle” [1, 2].The high-energy phosphate bond istransferred from ATP to creatine atthe site of ATP production (i.e. themitochondria), yielding phosphocrea-tine and ADP. This reaction is cataly-zed by the mitochondrial creatinekinase isoenzyme. Phosphocreatine,which is a smaller molecule than ATP,then diffuses through the cytoplasmto the site of ATP utilization, themyofibrils, where the back reactionoccurs, ATP is reformed and is usedfor contraction. This reaction iscatalyzed by the myofibrillar-boundMM-creatine kinase isoenzyme. Freecreatine then diffuses back to themitochondria. This “energy shuttle” is required, because the low freecytosolic ADP concentration (40-80 µM)does not provide the necessary capacity for back diffusion to themitochondria [1, 3], while freecreatine is present at concentrationsthat are at least two orders of mag-nitude greater. The second crucialcellular function of phosphocreatineand creatine kinase is to maintainfree cytosolic ADP at low concentra-tion. When ATP is hydrolysed, inorganicphosphate is formed. Inorganicphosphate increases when ATPutilization exceeds ATP production,for example during ischemia.

Experimental 31P-MR Spectroscopy31P-MRS has been a standard methodin experimental cardiology since the1970s [4]. A large body of literatureexists on experimental applicationsof 31P-MRS to the heart, and the

Introduction

Cardiac magnetic resonance (CMR)imaging (MRI) is an excellent methodfor obtaining anatomical and functio-nal information on the heart, butoffers little insight into the biochemi-cal state of cardiac tissue. In contrast,31P-MR spectroscopy (31P-MRS) is theonly available method for the non-invasive study of cardiac energymetabolism without need for theapplication of external radioactivetracers (as required in PET). Non-invasive imaging of cardiac metabo-lism is a long-standing Cardiologists’dream, and the main reason why 31P-MRS has not yet fulfilled its promisein clinical Cardiology is the limitedresolution of the method: the MRsensitivity of the 31P-nucleus is 6.6%that of 1H. Even more significantly, inthe heart, 31P-containing metabolites[(adenosinetriphosphate (ATP),phosphocreatine (PCr), inorganicphosphate (Pi)] are present in con-centrations that are several orders ofmagnitude lower than those of 1Hnuclei of water and fat (1-15 mM vs.110 M). Thus, the signals we aim tointerrogate in 31P-MRS are ~6 ordersof magnitude weaker than thoseused in 1H-MRI.

Cardiac Energetics31P-MRS is suitable for the repeatedmeasurement of cardiac high-energyphosphate metabolism, to non-invasively monitor the energeticstate of the heart. ATP is the only



Figure 1 31P-MR spectra from anuntreated perfused rat heart and aheart treated with Verapamil duringcontrol, at the end of 30 min. ofhypoxia and at the end of 30 min. ofreperfusion. NAD = Nicotine adeninedinucleotide. See the depletion of phosphocreatine and ATP duringhypoxia, and the recovery of phos-phocreatine, but not of ATP, duringreoxygenation in the untreatedheart. These changes of cardiacenergetics are attenuated by thepresence of the Ca++ antagonistVerapamil. CP or CrP = creatinephosphate, a term synonymous withphosphocreatine.

Figure 2 31P-MR spectrum from thehuman heart of a volunteer and of apatient with dilated cardiomyopathy.The reduction of the phosphocreatineresonance in the patient is apparent.From Neubauer S et al. Eur Heart J 1995;16(Suppl O):115-118. Reproduced with permissionfrom European Society of Cardiology.

concentrations are evaluated bycomparing the tissue 31P-resonanceareas to those of an external 31P-reference standard. 31P-MRS alsoallows the quantification of intracel-lular pH (pHi), from the chemical shiftdifference between phosphocreatineand inorganic phosphate, which ispH-sensitive. Because the method isnon-invasive, spectra can be acquiredsequentially, and the dynamicresponse of energy metabolites toischemia, hypoxia or inotropicstimulation can be followed in oneexperiment (Fig. 1).

Clinical Cardiac 31P-MRS Studies

Methodological aspectsClinical cardiac 31P-MR spectroscopyfaces major technical challenges:total examination time should not bemore than one hour, and the time forsignal acquisition is thereby limited.The heart is a rapidly moving organ,requiring gating to the heart beat,and, when resolution is furtherimproved, ultimately to respiration aswell [6]. The cardiac muscle liesbehind the chest wall skeletal muscle

which by itself creates a strong 31P-signal that requires suppression. Thisnecessitates the use of localizationtechniques such as DRESS (depth-resolved surface coil spectroscopy),rotating frame, 1D-CSI (chemicalshift imaging), ISIS (image-selectedin vivo spectroscopy), and 3D-CSI,which are described in detail else-where [7]. For most spectroscopictechniques, 1H scout images are firstobtained, which are used to selectthe spectroscopic volume(s). The lowsensitivity of 31P-MRS has requiredlarge voxel sizes, typically ~30 ml. A

Untreated

Volunteer Severe DCM

Control

γ–ATP

α–ATP

γ–ATP α–ATP

β–ATP

β–ATP

NAD

CP

PCr

PDE2.3-DPG

Pi

MPEHypoxia

ppm10

10 5 0 –5 –10 –15 –20 10 5 0 –5 –10 –15 –20

0 –10 –20 10 0 –10 –20

Reoxygenation

Verapamil 10 -5M

recovery [12]. After revascularization,phosphocreatine/ATP ratios no longerchanged with exercise. If high spatial(1-5 ml) and temporal (stress duration<15 min.) resolution for this approachcould be achieved, then this methodmight well find its way into routinecardiology. A 31P-MRS stress testwould, for example, allow the non-invasive study of the efficiency ofrevascularization procedures or ofvarious anti-anginal therapies. Pohost’sgroup [13] recently reported on thepathophysiologic mechanisms ofexercise-induced chest pain in womenwith normal coronary arteries: in 7/35 women with chest pain andnormal coronary arteries, the phos-phocreatine/ATP ratio decreased by29 ± 5% during hand grip exercise.These findings provide direct evidenceof exercise-induced myocardialischemia in women with chest painand normal coronary arteries.

Assessment of Myocardial Viability.Both biochemical [14] and 31P-MRSmeasurements [15, 16] (Fig. 4) inanimal models have shown thatmyocardial scar tissue containsnegligible amounts of ATP (<1% ofnormal levels), while in stunned andhibernating [17] myocardium, ATPlevels remain close to normal. There-fore, a non-invasive method thatallows measurement of myocardialATP levels with high spatial (1-5 ml)and acceptable temporal resolution(<30 min.) should be quite suitablefor evaluation of myocardial viability.Only a few clinical studies havepreviously addressed this issue [18,19]. For example, Kalil-Filho et al.studied 29 patients with anteriorinfarction 4 and 39 days after MI. Allpatients showed akinetic anteriormyocardium, which had recoveredfunction at the time of the secondexamination. Phosphocreatine/ATPratios were normal in stunnedmyocardium (1.51 +/- 0.17 vs. 1.61



It has been impossible up to nowto interrogate the posterior wall ofthe human heart due to its distancefrom the 31P-surface coil. However,Pohmann and von Kienlin [11] haverecently implemented acquisition-weighted 31P-chemical shift imagingin volunteers at 2T. Acquisition-weighting significantly reduced thesignal contamination betweenadjacent voxels, and it has beenpossible to obtain human heart ATPand phosphocreatine images with 16 ml spatial resolution within 30 min.Furthermore, for the first time, 31P-spectra from the posterior wallcould be obtained.

In a normal human heart, thephosphocreatine/ATP ratio is ~1.8 [7]although, due to different methodsbeing used, the range of “normal”phosphocreatine/ATP ratios reportedin the literature is considerable, fromabout 1.1 to 2.4.

Ischemic Heart DiseaseThere are, in principle, two appli-cations of 31P-MRS in ischemic heartdisease:“Biochemical Ergometry” for Detec-tion of Exercise-Induced Ischemia.A decrease of phosphocreatine andan increase of inorganic phosphateare amongst the very earliest meta-bolic responses in myocardial ische-mia. Thus, a “biochemical stress test”,able to measure these metabolites in human myocardium with hightemporal and spatial resolution, wouldallow detection of the regional bio-chemical consequences of myocardialischemia at rest, during exercise and recovery. The principal feasibilityof this approach has been demon-strated: Weiss et al. showed that inpatients with high grade LAD stenosis,phosphocreatine/ATP ratios werenormal at rest; phosphocreatine/ATPdecreased during hand grip exercisefrom 1.5 ± 0.3 to 0.9 ± 0.2, andreturned towards normal during

typical 31P-MR spectrum of a healthyvolunteer is shown in Figure 2.Compared to the rat heart spectrum,two additional resonances aredetected: 2,3-diphosphoglycerate(2,3-DPG), due to the presence ofblood (erythrocytes) in the inter-rogated voxel, and phosphodiesters(PDE), a signal due to membrane as well as serum phospholipids. The2,3-diphosphoglycerate resonancesoverlap with the inorganic phosphatepeak, which therefore cannot bedetected in blood-contaminatedhuman 31P-MR spectra. For relativequantification of human 31P-spectra,the phosphocreatine/ATP ratio iscalculated. Phosphocreatine/ATP isan index of the energetic state of the heart, because any imbalancebetween oxygen supply and demandwill lead to substantial depletion ofphosphocreatine before ATP levelsstart to decline (due to the CK equili-brium constant). For area integrationof resonances in human 31P-spectra, a Lorentzian line fitting algorithm isapplied.

Absolute quantification of phos-phocreatine and ATP is technicallydemanding, but is clearly a necessityin the long term, as the phosphocrea-tine/ATP ratio cannot detect simul-taneous decreases of both phospho-creatine and ATP that occur, forexample, in the failing [8] or in theinfarcted myocardium [9]. The most sophisticated method for this isSLOOP (“spectral localisation withoptimum pointspread function”),which allows for curved regions ofinterest and absolute quantificationwith high accuracy [10]. SLOOPrequires use of a 31P reference stan-dard, flip angle calibration, precise B1

field mapping and determination ofmyocardial mass in the interrogatedvoxel (Fig. 3). However, almost allprevious clinical 31P-MRS studies haveused the phosphocreatine/ATP ratiofor quantification.


Figure 3 SLOOP MR spectroscopy. a) a: B1-corrected proton short-axisimage of the heart as used for seg-mentation. A threshold filter wasapplied to cut off the noise in theimage regions distant from the coil.After segmentation, the matrix sizewas reduced to limit computationalburden. The contours of this compart-ment map with decreased resolutionare overlaid. b: For the same slice thespatial response function (SRF) of theleft ventricular myocardium is over-laid with the original compartmentborders. c: The color encoding schemefor the complex value of the SRF. b) 31P-spectra from the left ventricularwall, obtained from SLOOP recon-struction of 3D-CSI datasets, acquiredwith NOE (top), and without (bottom),and calculated with T1 and the off-resonance effect of PCr. These spectrawere reconstructed from the localizedtime domain signals using zero-fillingto 2048 data points and an exponen-tial filter with a line-broadening of 7 Hz prior to fast Fourier transform.

Figure 4 Central slices of a 3D-31P-chemical shift imaging experiment inan isolated, chronically infarcted ratheart. 12 Tesla, nominal voxel size54 µl. On the 1H image (upper left),the infarcted area in the anteriorwall (a) appears just barely brighterthan the unaffected myocardium. In 31P-images, however, the infarctedarea can be identified due to thecomplete absence of phosphocreatine(PCr; upper right) and of g-ATP(lower left). Inorganic phosphatecannot be detected due to its lowconcentration (lower right).

a

a

b c

b

γ–ATP Pi

1H-Image PCr

α–ATP β–ATP

PCr

γ–ATP

10ppm

Im(SRF)

Re(SRF)

0 –10 –20


energy-costly, such as beta-receptormimetic or phosphodiesterase inhibi-tor drugs, while increasing cardiacoutput in the short term, has ultima-tely increased mortality, while anytreatment that is energy-sparing,such as betablockers, ACE-inhibitors,or Angiotensin-II-receptor blockers,has improved survival in long-termstudies. For clinical heart failure trials,it would thus be highly attractive tomonitor the early energetic responseof the myocardium to new classes ofagents for the treatment of heartfailure, and it is conceivable that thephosphocreatine/ATP ratio or absoluteconcentrations of phosphocreatineand ATP may be powerful surrogateparameters for mortality in heartfailure trials. Our initial observationwas that in 6 patients with dilatedcardiomyopathy treated with stan-dard medical therapy including ACEinhibitors, digitalis, diuretics and, in 4 patients, beta-blockers, clinicalrecompensation occurred over 3 months. Over this time, the phos-phocreatine/ATP ratio of the patientsimproved significantly from 1.51 ±0.32 to 2.15 ± 0.27 [25]. Apart fromthis early anecdotal evidence, nosystematic study has been publishedso far using 31P-MRS to monitorcardiac energetics during heartfailure treatment and such trials areeagerly awaited.

In line with the concept thatenergy-sparing forms of therapyimprove prognosis in heart failure, isour observation that phosphocreatine/ATP ratios hold prognostic informa-tion on survival of patients with heartfailure that extends beyond theprognostic relevance of clinical andhemodynamic variables. We showedthat in dilated cardiomyopathy, themyocardial phosphocreatine/ATPratio was a better predictor of long-term survival than left ventricularejection fraction or New York HeartAssociation (NYHA) class [27] (Fig. 6).



Heart FailureExperimental studies have universallyshown reduced phosphocreatinelevels in chronically failing myocar-dium (e.g. [14, 23]). Hardy et al. firstdemonstrated that the myocardialphosphocreatine/ATP ratio is signifi-cantly reduced (from 1.80 ± 0.06 to1.46 ± 0.07) in symptomatic patientswith heart failure of ischemic or non-ischemic origin [24]. We reportedthat the decrease of phosphocreatine/ATP ratios in dilated cardiomyopathycorrelated with the clinical severity of heart failure, according to the NewYork Heart Association (NYHA) class[25] and also with left ventricularejection fraction [26]. Thus, phos-phocreatine/ATP ratios decrease forthe more symptomatic stages ofheart failure. However, in heartfailure, both phosphocreatine andATP levels decrease in parallel [8],and this cannot be detected bymeasurement of phosphocreatine/ATP ratios. Using the SLOOP techni-que for absolute quantification, wehave recently studied patients withheart failure due to dilated cardiomy-opathy (ejection fraction 18%), andfound that absolute phosphocreatinelevels were reduced by 51%, ATPlevels by 35%, while the phospho-creatine/ATP ratio decreased by 25%only [21] (Fig. 5). Thus, the trueextent of changes in energy meta-bolism in heart failure is underesti-mated when phosphocreatine/ATPratios are measured rather thanabsolute concentrations. Therefore,measurement of absolute concen-trations of high-energy phosphatesshould become the standard approachfor human 31P-MRS studies in heartfailure as well as for other cardiacpathologies.

The energy-deprivation hypothesisin heart failure has recently gainedfresh support, since an analysis ofheart failure trials from the past decadeshows that any treatment that is

+/- 0.18 in volunteers) and did notchange during the recovery of con-tractile function (1.51 +/- 0.17 vs.1.53 +/- 0.17). However, loss ofmyocardial tissue, such as followinga non-viable infarct leading to necro-sis and scar formation, primarilyleads to a reduction of both phospho-creatine and ATP, and detection ofthis requires measurement of absoluteconcentrations of high-energy phos-phates. Feasibility for this approachwas shown by Yabe et al. [9]: absolutemyocardial ATP content was signifi-cantly reduced in patients with fixedthallium defects (non-viable) but wasunchanged in patients with reversibledefects (viable myocardium). Whilethese initial results are promising,technical advances are required toachieve the necessary resolution tomake 31P-MRS evaluation of viabilityclinically practicable.

HypertensionVery few studies have examinedpatients with left ventricular hyper-trophy due to chronic hypertension.Lamb et al. demonstrated reducedphosphocreatine/ATP ratios in patientswith hypertension both at rest andduring dobutamine-stress testing[20]. Furthermore, phosphocreatine/ATP ratios correlated inversely withindices of diastolic function (E dece-leration peak). In contrast, anotherstudy showed no significant changesof cardiac energetics in hypertension[21]. It is likely that differences inpatient characteristics, such asseverity and duration of hypertension,are responsible for this discrepancy,since experimental data clearlysuggest that cardiac energetics areimpaired in longstanding hyperten-sion [22]. In advanced hypertensiveheart disease, 31P-MRS should allowthe energetic correlates of hyper-trophy regression during variousforms of antihypertensive therapy to be followed.


Thus, 31P-MRS may become a tool forprognosis evaluation in heart failure.

Diabetic CardiomyopathyMost recently, 31P-MRS has beenapplied to the study of the uniquemetabolic alterations present indiabetic cardiomyopathy, whichdevelops on the basis of combinedhyperglycaemia and increasedplasma free fatty acids. Whereascardiac glucose uptake is reduceddue to insulin resistance, uptake offree fatty acids and their �-oxidationwithin the mitochondrium is increa-sed. This further inhibits glucoseuptake, glycolytic ATP synthesis andthe generation of glucose derivedAcetyl-CoA. Additionally, �-oxidationof free fatty acids increases theexpression of uncoupling proteins,which shunt protons away from therespiratory (electron transport) chain,leading to reduced ATP generation.Scheuermann-Freestone et al. reportedrecently that myocardial phospho-creatine/ATP ratios are substantiallyreduced in patients with type IIdiabetes in the presence of maintainedleft ventricular function [28], andthat phosphocreatine/ATP ratioscorrelate with acute plasma levels offree fatty acids and glucose (Fig. 7).Diamant et al [29] confirmed thesefindings and also demonstrated thephosphocreatine/ATP ratios correla-ted with indices of diastolic function.Metzler’s group [30] demonstratedsimilar reductions of phosphocreatine/ATP ratios in otherwise normal type Idiabetics. Together, these findingssuggest that altered cardiac energe-tics contribute to the development ofdiabetic cardiomyopathy.

Genetic Heart DiseaseOne of the most promising areas forclinical cardiac 31P-MRS is the non-invasive phenotyping of cardiomyo-pathies due to specific gene defects.While it is still early days, it is conceiv-

Figure 6 Kaplan-Meier life table analysis for total mortality ofdilated cardiomyopathy patients divided into two groups split by themyocardial phosphocreatine/ATP ratio (<1.60 vs. >1.60). Patientswith an initially low PCr/ATP ratio showed increased mortality overthe study period of, on average, 2.5 years.

Figure 5 Absolute and relative concentrations of phosphocreatine(PCr) and adenosine triphosphate (ATP) in volunteers (VOL) and inpatients with dilated cardiomyopathy (DCM). Individual as well asmean +- SD values are shown for each group. PCr: VOL vs. DCM p <0.000001; ATP: VOL vs. DCM p < 0.0003; PCr/ATP ratio: VOL vs. DCM p = 0.06. Reproduced with permission from Journal of theAmerican College of Cardiology.

[PCr] [ATP] PCr/ATP

PCr/ATP < 1.60

p = 0.036

PCr/ATP > 1.60

VOL

mm

ol/k

g w

et w

eigh

t

Tota

l Mor

talit

y (%

)

Days of follow-up

mm

ol/k

g w

et w

eigh

t

DCM VOL DCM VOL DCM


0 500

50

40

30

20

10

0

1000 1500



able that specific gene defects mayeventually be identifiable by a specificmetabolic profile, as detected non-invasively by MRS. Most work in thisarea has so far concentrated onhypertrophic cardiomyopathy (HCM).The Watkins group has recentlyproposed a new energy depletionparadigm as the one unifying patho-physiological mechanism underlyingHCM due to various gene mutations[31]. Human studies in HCM haveshown reduced phosphocreatine/ATPratios in myocardial tissue affectedby hypertrophy. Jung et al. [32]demonstrated that young, asympto-matic patients with HCM show asignificantly reduced phosphocreati-ne/ATP ratio, indicating that energe-tic imbalance occurs at an early stageof the disease process. Crilley et al.showed reduced phosphocreatine/ATP ratios in patients with threedifferent mutations causing HCM[33]. Importantly, the phosphocrea-tine/ATP ratio was reduced whetherhypertrophy was present or not,suggesting that hypertrophy is notper se the primary mechanism redu-cing phosphocreatine/ATP ratios inHCM. In the future, large patientcohorts with HCM and identifiedspecific gene defects will have to bestudied to test the hypothesis thatmetabolic phenotyping using 31P-MRScan non-invasively predict specificgenetic mutations in individual HCMpatients.

Summary and FuturePerspective31P-MR spectroscopy is a fascinatingtechnique, allowing insight intocardiac energy metabolism in normaland diseased heart. The main obstaclesfor routine implementation of cardiac31P-MRS are its technical complexityand limited resolution. Thus, a majortechnological effort is required toimprove spatial and temporal resolu-

Figure 7 a) Typical cardiac 31P-MRspectra from a normal control (upperspectrum) and a patient with type 2diabetes (lower spectrum), showinglower PCr/ATP ratio in the patient.2,3-DGP indicates 2,3-diphosphogly-cerate; PDE, phosphodiesters; PCr,phosphocreatine; a, b, and g indica-te the three phosphate groups ofATP. b) Cardiac PCr/ATP ratios corre-lated negatively with the plasma freefatty acid (FFA) concentrations (r2 =0.32, P = 0.01) for all subjects andcorrelated positively with the plasmaglucose concentrations (r2 = 0.55, P = 0.05) for the patients with type 2diabetes, but there was no corre-lation for control subjects.

Figure 7a

Figure 7b

tion. This will include advances in coiland sequence design, and implemen-tation of MRS at higher field strength [34, 35]. In addition, highsignal/noise spectra need to beacquired so that measurementvariability is reduced, and standardi-zed acquisition and quantificationprotocols will have to be developed,so that the method produces measu-rements that are robust and univer-sally accepted. If these development

goals can be achieved, high resolutionmetabolic imaging of the heart maybecome part of standard diagnosticpathways in Cardiology.

References

[ 1 ] Wallimann T, Wyss M, Brdiczka D, Nicolay K,Eppenberger HM. Intracellular compartmen-tation, structure and function of creatine kinaseisoenzymes in tissues with high and fluctuatingenergy demands: The ‘phosphocreatine circuit’for cellular energy homeostasis. Biophys J. 1992; 281:21-40.

αβ

ATP

ppm

ppm

2.3-DPG

PDE

Diabetic patient

PCr/

ATP

[Free fatty acid] (mM) [Glucose] (mM)

� Controls� Diabetics

r2 = 0.32p < 0.01

r2 = 0.55p < 0.05

ControlPCr

γ

10 5 0 –5 –10 –15 –20

10 5 0 –5 –10 –15 –20

10

3

2

1

0

0.5 1 5 10 15 20



[ 2 ] Ingwall JS, Kramer MF, Fifer MA, Lorell BH, Shemin R, Grossman W, Allen PD. The creatine kinase system in normal anddiseased human myocardium. N Engl J Med. 1985; 313:1050-4.

[ 3 ] Jacobus WE. Respiratory control and theintegration of heart high-energy phosphatemetabolism by mitochondrial creatine kinase.Annu Rev Physiol. 1985; 47:707-25.

[ 4 ] Garlick PB, Radda GK, Seeley PJ. Phosphorus NMR studies on perfused heart.Biochem Biophys Res Commun. 1977; 74:1256-62.

[ 5 ] Neubauer S. Cardiac magnetic resonancespectroscopy: potential clinical applications.Herz. 2000; 25:452-60.

[ 6 ] Kozerke S, Schar M, Lamb HJ, Boesiger P.Volume tracking cardiac 31P spectroscopy.Magn Reson Med. 2002; 48:380-4.

[ 7 ] Bottomley PA. MR spectroscopy of thehuman heart: the status and the challenges.Radiology. 1994; 191:593-612.

[ 8 ] Shen W, Asai K, Uechi M, Mathier MA,Shannon RP, Vatner SF, Ingwall JS. Progressive loss of myocardial ATP due to a lossof total purines during the development ofheart failure in dogs: a compensatory role forthe parallel loss of creatine. Circulation. 1999;100:2113-8.

[ 9 ] Yabe T, Mitsunami K, Inubushi T, Kinoshita M. Quantitative measurements ofcardiac phosphorus metabolites in coronaryartery disease by 31P magnetic resonancespectroscopy [see comments]. Circulation. 1995; 92:15-23.

[ 10 ] Meininger M, Landschutz W, Beer M,Seyfarth T, Horn M, Pabst T, Haase A, Hahn D,Neubauer S, von Kienlin M. Concentrations of human cardiac phosphorusmetabolites determined by SLOOP 31P NMRspectroscopy. Magn Reson Med. 1999; 41:657-63.

[ 11 ] Pohmann R, von Kienlin M. Accurate phosphorus metabolite images of thehuman heart by 3D acquisition-weighted CSI. Magn Reson Med. 2001; 45:817-26.

[ 12 ] Weiss RG, Bottomley PA, Hardy CJ,Gerstenblith G. Regional myocardial meta-bolism of high-energy phosphates duringisometric exercise in patients with coronaryartery disease [see comments]. N Engl J Med. 1990; 323:1593-600.

[ 13 ] Buchthal SD, den Hollander JA, Merz CN,Rogers WJ, Pepine CJ, Reichek N, Sharaf BL, Reis S, Kelsey SF, Pohost GM. Abnormal myocardial phosphorus-31 nuclear magnetic resonance spectroscopy inwomen with chest pain but normal coronaryangiograms. N Engl J Med. 2000; 342:829-35.

[ 14 ] Neubauer S, Horn M, Naumann A, Tian R, Hu K, Laser M, Friedrich J, Gaudron P,Schnackerz K, Ingwall JS, et al. Impairment of energy metabolism in intact residual myocardium of rat hearts with chronic myo-

cardial infarction. J Clin Invest. 1995; 95:1092-100.

[ 15 ] Friedrich J, Apstein CS, Ingwall JS. 31Pnuclear magnetic resonance spectroscopicimaging of regions of remodeled myocardiumin the infarcted rat heart. Circulation. 1995; 92:3527-38.

[ 16 ] von Kienlin M, Rosch C, Le Fur Y, Behr W,Roder F, Haase A, Horn M, Illing B, Hu K, Ertl G,Neubauer S. Three-dimensional 31P magnetic resonance spectroscopic imaging of regional high-energy phosphate metabolism in injured rat heart. Magn Reson Med. 1998; 39:731-41.

[ 17 ] Flameng W, Vanhaecke J, Van Belle H,Borgers M, De Beer L, Minten J. Relationbetween coronary artery stenosis and myocardial purine metabolism, histology andregional function in humans. J Am Coll Cardiol. 1987; 9:1235-1242.

[ 18 ] Kalil-Filho R, de Albuquerque CP, Weiss RG, Mocelim A, Bellotti G, Cerri G, Pileggi F. Normal high-energy phosphate ratiosin stunned human myocardium. J Am Coll Cardiol. 1997; 30:1228-1232.

[ 19 ] Beer M, Sandstede J, Landschutz W,Viehrig M, Harre K, Horn M, Meininger M, Pabst T, Kenn W, Haase A, von Kienlin M,Neubauer S, Hahn D. Altered energy meta-bolism after myocardial infarction assessed by 31P-MR-spectroscopy in humans. Eur Radiol. 2000;10:1323-8.

[ 20 ] Lamb HJ, Beyerbacht HP, van der Laarse A, Stoel BC, Doornbos J, van der Wall EE,de Roos A. Diastolic Dysfunction in HypertensiveHeart Disease Is Associated With AlteredMyocardial Metabolism. Circulation. 1999; 99:2261-2267.

[ 21 ] Beer M, Sandstede J, Landschütz W,Seyfarth T, Lipke C, Köstler H, Pabst T, Kenn W,Meininger M, von Kienlin M, Horn M, Harre K,Hahn D, Neubauer S. Absolute concentrationsof myocardial high-energy phosphate meta-bolites in normal, hypertrophied and failinghuman myocardium, measured non-invasivelywith 31P-SLOOP-magnetic resonance spectroscopy. J Am Coll Cardiol. 2002; 40:1267-74.

[ 22 ] Perings SM, Schulze K, Decking U, Kelm M, Strauer BE. Age-related decline ofPCr/ATP-ratio in progressively hypertrophiedhearts of spontaneously hypertensive rats. Heart Vessels. 2000;15:197-202.

[ 23 ] Nascimben L, Friedrich J, Liao R, Pauletto P, Pessina AC, Ingwall JS. Enalapriltreatment increases cardiac performance andenergy reserve via the creatine kinase reactionin myocardium of Syrian myopathic hamsterswith advanced heart failure. Circulation. 1995; 91:1824-33.

[ 24 ] Hardy CJ, Weiss RG, Bottomley PA,Gerstenblith G. Altered myocardial high-energyphosphate metabolites in patients with dilatedcardiomyopathy. Am Heart J. 1991;122:795-801.

[ 25 ] Neubauer S, Krahe T, Schindler R, Horn M, Hillenbrand H, Entzeroth C, Mader H,Kromer EP, Riegger GA, Lackner K, Ertl G. 31P magnetic resonance spectroscopy indilated cardiomyopathy and coronary arterydisease. Altered cardiac high-energy phosphatemetabolism in heart failure. Circulation. 1992; 86:1810-8.

[ 26 ] Neubauer S, Horn M, Pabst T, Gödde M,Lübke D, Illing B, Hahn D, Ertl G. Contributionsof 31P-magnetic resonance spectroscopy to theunderstanding of dilated heart muscle disease.Eur Heart J. 1995;16 (Suppl O):115-118.

[ 27 ] Neubauer S, Horn M, Cramer M, Harre K,Newell JB, Peters W, Pabst T, Ertl G, Hahn D,Ingwall JS, Kochsiek K. Myocardial phospho-creatine-to-ATP ratio is a predictor of mortality in patients with dilated cardiomyopathy. Circulation. 1997; 96:2190-6.

[ 28 ] Scheuermann-Freestone M, Madsen PL,Manners D, Blamire AM, Buckingham RE, Styles P, Radda GK, Neubauer S, Clarke K. Abnormal cardiac and skeletal muscle energymetabolism in patients with type 2 diabetes. Circulation. 2003; 107:3040-6.

[ 29 ] Diamant M, Lamb HJ, Groenevelt Y, et al.Diastolic dysfunction is associated with alteredmyocardial metabolism in asymptomaticnormotensive patients with well-controlled typeII diabetes mellitus. J Am Colll Cardiol. 2003; 42:328-35.

[ 30 ] Metzler B, Schocke MF, Steinboeck P,Wolf C, Judmaier W, Lechleitner M, Lukas P,Pachinger O. Decreased high-energy phosphateratios in the myocardium of men with diabetesmellitus type I. J Cardiovasc Magn Reson. 2002; 4:493-502.

[ 31 ] Ashrafian H, Redwood C, Blair E, Watkins H. Hypertrophic cardiomyopathy: a paradigm for myocardial energy depletion. Trends Genet. 2003;19:263-8.

[ 32 ] Jung WI, Sieverding L, Breuer J, Hoess T, Widmaier S, Schmidt O, Bunse M, van Erckelens F, Apitz J, Lutz O, Dietze GJ. 31P NMR spectroscopy detects metabolicabnormalities in asymptomatic patients withhypertrophic cardiomyopathy. Circulation. 1998; 97:2536-2542.

[ 33 ] Crilley JG, Boehm EA, Blair E, Rajago-palan B, Blamire AM, Styles P, McKenna WJ,Ostman-Smith I, Clarke K, Watkins H. Hypertrophic cardiomyopathy due to sarcomericgene mutations is characterized by impairedenergy metabolism irrespective of the degree of hypertrophy. J Am Coll Cardiol. 2003; 41:1776-82.

[ 34 ] Lee RF, Giaquinto R, Constantinides C,Souza S, Weiss RG, Bottomley PA. A broadbandphased-array system for direct phosphorus andsodium metabolic MRI on a clinical scanner.Magn Reson Med. 2000; 43:269-77.

[ 35 ] Hetherington HP, Luney DJ, Vaughan JT,Pan JW, Ponder SL, Tschendel O, Twieg DB,Pohost GM. 3D 31P spectroscopic imaging ofthe human heart at 4.1 T. Magn Reson Med. 1995; 33:427-31.

FOV: 240x240x200 mm3.

Spatial resolution: 2.1x2.1x3.0 cm3 = 13 ml.

ECG-gating, total exam time: around 45 minutes with positioningand acquisition of proton referenceimages.

Post-processing: institute-developed3D-CSI software.

Date and Location of Scan: 04-22-2002 CEMEREM Marseilles,France.

Scanner Model: MAGNETOM VisionPlus at 1.5T with broadband option.

Coil used: Siemens 31P-heart/liversurface coil.

Software Version: Numaris 3.5 VB 33.

Discussion

3D acquisition-weighted CSI hasenabled the acquisition of regionalspectra with good localization quality– here demonstrated by a PCr meta-bolic image. The spectra show meta-bolic differences between differentlocations around the necrotic area inthe apex. CSI here gives non-invasiveand direct access to metabolism. The spatial resolution achieved in themetabolic map is low compared with conventional MRI, but close toroutinely practised SPECT imageresolutions. Metabolic mapping ofHEP might enable localization ofjeopardized tissue and help in thedecision whether or not to revascu-larize in case of severe coronarystenosis.



Frank Kober Ph.D.J. Quilici, E. Guedj, T. Caus, J.L. Bonnet, P.J. Cozzone, M. Bernard

Centre d’Exploration Métaboliquepar Résonance Magnétique (CEMEREM)Marseilles, France

31P-Chemical Shift Imaging for MyocardialInfarction

lite images and spectra of a patientwith anterior wall infarction. All MRexperiments were conducted on aSiemens Vision Plus system operatingat 1.5T.

Spectra and Images

The figure shows a short-axis protonturbo-FLASH image and a spatiallycorresponding PCr image of a patient(70 years, 84 kg) with apical anteriorwall infarction (small antero-apicalnecrosis) after LAD occlusion. Spectrataken from two different regions inthe myocardium are shown. Thepatient has an akinetic anterior wall.The shape drawn in red and purplerepresents the voxel contours at 64and 33 percent maximum of thepoint spread function at the locationwhere the spectra were obtained andthat enables an estimate to be madeof the localization quality. The pres-ented image slice is located in themiddle part of the left ventricle thatdoes not contain necrotic tissue. The spectra, however, show a visiblylower PCr/ATP ratio in the anteriorwall than in the inferior wall, indica-ting metabolic alteration in viabletissue adjacent to the necrosis. ThePCr image demonstrates the locali-zation quality achieved with acquisi-tion-weighted 3D-CSI.

Sequence Details

CEMEREM-developed implemen-tation derived from factory 31P 3D-CSIsequence.

3D Hanning acquisition-weighted 31P chemical shift imaging withpulsed NOE.

2228 phase encoding steps acquiredin radial order, four accumulations inthe center of k-space.

Introduction

Phosphorus-31 magnetic resonancespectroscopy (31P-MRS) gives non-invasive access to the concentrationsof high-energy phosphates (HEP).The ratio of the concentrations ofATP and phosphocreatine (PCr) is awell-established marker of cardiacand muscular ischemia. In the future, 31P-MRS could complement the set of parameters measured by othertechniques in order to improve theregional characterization of myo-cardial tissue. In particular, it couldcontribute to an early prediction offunctional recovery after surgicalintervention in myocardial infarction.

Clinical Method Description

3D 31P chemical shift imaging (CSI)provides spatially localized informa-tion of the entire heart in one singleexam, allowing the selection ofregions of interest for spectral evalu-ation after completion of the exam.In the past, the localization quality of CSI techniques was substantiallyaffected by contamination of thespectra from surrounding tissue(skeletal muscle, liver) and blood,making human studies difficult andprone to errors. Acquisition-weightedCSI helps to partly overcome theproblem of spectral contaminationand therefore to increase the sensi-tivity of the experiment. Moreover,3D-CSI enables the concentration ofHEP metabolites to be mappedspatially. We present the PCr metabo-



PCr

anterior wall

AT

PCr

inferior wall

AT

spectroscopy combined with surfaceelectromyography. A paragraphcovers the issue of functional investi-gations of muscle using MR Imaging(MRI). For each paragraph, a shortsection will be devoted to technolo-gical issues.

II. Combined investigations of muscle fatigue using MR spectroscopy and surfaceelectromyographyMuscle fatigue is usually defined asthe loss of maximum force-generatingcapacity and may occur at the varioussites along the pathway from thecentral nervous system through tothe intramuscular contractile machi-nery. As well as core factors [6], thereare peripheral factors that couldinterfere with force production,including metabolic inhibition of thecontractile process and E-C couplingfailure [7; 8]. Therefore, isolated meta-bolic investigations offer a limitedwindow towards the investigation ofperipheral factors and a few groupshave reported combined investiga-tions of electrical and metabolicchanges associated with the failureof muscle force production in orderto widen the experimental field ofresearch related to fatigue.

Electromyographic components offatigue: During the transition fromrest to exercise, typical electromyo-graphic changes, such as a rise in theintegrated EMG (iEMG) and a shift of the power spectrum towards low


CLINICALMRS MULTI NUCLEI

David BENDAHAN, M.D., Benoît GIANNESINI, M.D.,Patrick J. COZZONE, Ph.D.

Centre de Résonance MagnétiqueBiologique et Médicale (CRMBM)

Faculté de Médecine de Marseilles,Marseilles, France

Functional investigation of exercisingmuscle: A non-invasive MagneticResonance Spectroscopy – MagneticResonance Imaging approach

Summary

Magnetic Resonance Spectroscopyand Imaging can be used in order toinvestigate muscle function non-invasively. Most of the hypothesesrelated to muscle fatigue – definedas the decline in muscle performanceduring exercise – have been elabora-ted on the basis of experimentalresults obtained in vitro and theirphysiological relevance has neverbeen clearly demonstrated in vivo. Inthat context, non-invasive methodssuch as 31-phosphorus magneticresonance spectroscopy (31P MRS)and surface electromyography (EMG)have been used to understand bothmetabolic and electrical aspects ofmuscle fatigue under physiologicalconditions. However, muscle activa-tion during exercise can be mappedin terms of both localisation andintensity, using MR Imaging and moreparticularly T2 changes.

The main results related to bothissues, i.e. the metabolic and electri-cal aspects of fatigue and the MRIfunctional investigation of exercisingmuscle, are discussed in this work. All experimental data were acquiredon a Siemens vision Plus systemoperating at 1.5 teslas.

Key words: skeletal muscle, fatigue,recruitment, MRS, MRI, energetics

I. Introduction

Functional non-invasive investigationof exercising muscle actually beganin 1974 when Hoult et al reportedthat high-energy phosphate com-pounds could be detected in vivousing 31-phosphorus (31P) magneticresonance spectroscopy (MRS) [1].Since then, a large number of studieshave been devoted to the investiga-tion of several metabolic aspects ofexercising muscle and so forth undera variety of conditions ranging frommuscular diseases to high-leveltraining (for review see [2-4]. However,the acute effects of exercise onmuscle MR images contrast were firstreported in 1988 by Fleckenstein etal. [5]. From then onwards, it hasbeen recognized that MRI could beused to distinguish between activeand non-active muscle groups,thereby providing some functionalassessments of exercising muscle.

In this report, we will describe inparticular the muscle fatigue investi-gated using magnetic resonance



Figure 1 Typical series of 31P MRspectra recorded on a Siemens VisionPlus system (surface coil) with a timeresolution of 2 s at end of exercise(first bottom spectrum) and duringthe following recovery period. Assignments of signals are specifiedon the top spectrum: ref: referencecompound, PME : phosphomonoe-sters, PDE : phosphodiesters, Pi:inorganic phosphate, PCr: phospho-creatine, a,b,g phosphate groups ofATP. Horizontal axis represents thechemical shifts expresses in Hz.Spectra are recorded from the thighmuscles.

frequencies, have been reported [9; 10]. While the iEMG rise mayindicate an increase in the firing rateof the motor unit’s discharge and/orrecruitment of additional musclefibers, the reasons accounting for theshift of the mean frequency of theEMG power spectrum remains amatter of debate. Alterations in thefrequency components of the [11]EMG power spectrum have beenreported during static and dynamiccontractions [12; 13] and variouslyrelated to accumulation of musclelactate [14] but not systematically tointracellular acidosis [10] and muscleconduction velocity [15]. Overall,metabolic causes have sometimesbeen put forward as reasons forsurface EMG (sEMG) alterations andcombined 31P MRS and sEMG investi-gations have been performed inorder to identify the link, if any.

Methodological issues: From amainly methodological point of view,this type of combined experiment isactually challenging. On the onehand, surface electrodes have to bepositioned beneath the musclesinvestigated and these could introduceadditional noise in MR spectra.Similarly, radiofrequency field inter-feres with sEMG signals. On the otherhand, the magnitudes of sEMGchanges are relatively small and thesignal must be amplified very close tothe source. In summary, technicallychallenging adaptations are necessaryin order to perform such combinedexperiments. Several types of solutionshave been proposed by a few researchgroups and convincing combinedmeasurements have been reported.

Main results: sEMG recordings canbe analyzed in the time domain (rootmean square) and in the frequencydomain (mean power frequency) andexercise-induced changes associatedwith both variables provide informa-

tion related to the amount andchronological activation of differentmotor units. When muscle contrac-tions are not performed voluntarily,electrical stimulations may be usedto acquire information regardingperipheral activation i.e. the excitabi-lity of the neuromuscular junctionand muscle membrane. In that case,supramaximal stimulations are usedand corresponding compound muscleaction potentials (CMAP or M-wave)are recorded. These variables – ormore exactly alterations of thesevariables – have been reportedduring fatiguing exercise and analyzedin the light of simultaneous meta-bolic changes such as depletion ofhigh-energy phosphates and accu-mulation of metabolic products, forexample ADP, Pi and H+ as determi-ned from 31P MR spectra recorded inexercising muscle (Fig. 1). Since thefirst study reported by Miller et al, a few combined analyses have beendevoted to the study of both meta-bolic and electrical changes linked tomuscle fatigue. As underlined below,the results are highly controversialand an unified position cannot beproposed at the moment. Miller’sresults were mainly based on theanalysis of the recovery phase follo-wing exercise and not on the exerciseperiod per se. They proposed at leastthree components of fatigue follo-wing the completion of exercise withone of them i.e. the recovery rate of MVC being associated with time-dependent changes in both PCr andintracellular pH [8]. They also reportedaltered M-wave (reduced in amplitudeand prolonged in duration) reflectingimpaired muscle membrane excita-tion and a decrease in neuromuscularefficiency (NME) that persists for alonger time as compared to both M-wave and metabolic changes.However, neither variable was closelyassociated with metabolic changes.Furthermore, the delayed recovery of

NME indicates a component ofimpaired muscular function that isindependent of high-energy phos-phates and intracellular pH. Inter-estingly, in both adductor pollicis andtibialis anterior muscles, low-intensityexercise produced a marked depres-sion of twitch tension with onlyminimal changes in MVC, M-wave,PCr and pH [16]. These observationsclearly indicate that for a low inten-sity-exercise, fatigue is largely due tochanges in E-C coupling with nocontribution of either central factors,impairment of the contractile mecha-nism, altered membrane propertiesor impaired neuromuscular transmis-sion [16]. However, the exact mecha-nism is poorly understood. Similarconclusions have been advanced byBendahan et al from combinedmeasurements performed duringisometric contractions of the forearmflexor muscles [17]. On the basis oftheir measurements, the existence of

cular junction and muscle membraneexcitability (using electrical stimula-tion) and muscle energetics (using31P MRS). Their main conclusion wasthat during a 4-min maximum iso-metric exercise involving the ankledorsiflexor muscles, central factorscontributed modestly (16%) tofatigue development. The remaining80% was apparently due to intramus-cular sources, primarily increasedproton concentration, given thatintracellular acidosis was significantlylinked to both the fall in MVC and theintegrated EMG decrease [25]. Therole of H+ accumulation in musclefatigue has often been considered asminor, but Kent-Braun’s study [21;25] and other studies [23; 26] havereported that a decline in force wasclosely related to it. Likewise, Milleret al observed such a relationship[26]. This strong association betweenfatigue and pH would be consistentwith the role of pH in feedback to thecentral nervous system and a sub-sequent alteration in central motordrive during the development offatigue. Similar measurements havebeen performed during an incremen-tal isometric exercise with the purposeof comparing the magnitude andmechanisms of ankle dorsiflexormuscle fatigue in young and oldersubjects [21]. From this study, theauthors’ main conclusion was thatyoung subjects fatigued more thanolder subjects regarding MVC measu-rements. However, at the conclusionof the exercise, there was a significantgender effect in that men had ahigher Pi and/or PCr (i.e. largerenergy consumption) than women.Meanwhile, intracellular pH fell moreand accumulation of Pi and/or H2PO4-were larger in the young comparedwith older subjects and in mencompared with women. Whateverthe group, a significant linear rela-tionship was found between H2PO4-and the fall in MVC regardless of the



links and the underlying mechanismscould differ among various groups of subjects [21]. Accumulation ofH2PO4- has been advocated as accounting for the loss of MVC butalso for the shift of the mean powerfrequency (MPF) of the EMG spec-trum [22; 23]. A quasi-linear decreasein mean power frequency (MPF) was found during an exhaustive calfmuscle exercise test and it wassignificantly correlated with H2PO4-concentration, which can be conside-red as resulting from both Pi accumu-lation and intramuscular acidosis. A similar inverse relationship hasbeen reported between musclelactate accumulation and the MPFshift in the vastus lateralis muscle,but surprisingly the expected linkwith intracellular pH was not obser-ved [10]. Bouissou et al also reportedthat systemic alkalosis was associa-ted with a greater spectrum shifttoward lower frequencies at exhaus-tion despite a level of muscle acidosissimilar to that in placebo conditionalthough muscle lactate concentra-tion was higher in alkalosis [10]. It is noteworthy that the shift in EMGpower spectrum toward lowerfrequencies can also account for theincreased RMS. When the lowfrequency component of the EMGsignal is increased, more myoelectricalsignals will be recorded since muscletissue and skin act as low-pass filters[24]. In addition, for submaximal(higher than 25% of MVC) exercise,the drop in MPF is independent offorce output [12].

More recently, combined analysesof central and peripheral contribu-tions to muscle fatigue have beenreported by Kent-Braun et al [21; 25].In order to quantify the respectivecontributions of central and peripheralfactors to fatigue development, theyperformed simultaneous non-invasivemeasurements of central activation(using electromyography), neuromus-

causal relationships between theextent of acidosis and EMG signs offatigue could not be firmly establis-hed. A shift of the median frequencyof the EMG power spectrum wasindeed observed as a sign of fatigue,but its time-dependent evolution wasnot linked at all to pH of PCr time-dependent changes given that EMGchanges characterizing fatigue wereonly noted beyond critical metabolicvalues [17]. Also of interest was theirobservation of a large scattering ofdata in a group of untrained subjectsdespite a standardization of theexercise protocol using the MVC [17].The effect of oxygen availability on a potential relationship betweenelectrical signs of fatigue and muscleenergetics alterations has also beeninvestigated [18]. Hypoxemia did notaffect the magnitude of metabolicchanges and the duration of contrac-tion. However, the rate of changes inintegrated sEMG was significantlymodified, thereby changing thecorrelated evolution of metabolicand electrical changes [18]. Thedownward shift of the relationshipsbetween myoelectrical and metabolicchanges under hypoxemia points tothe existence of a better E-C couplingand could indicate an adaptivemechanism [18].

However, a relationship betweenthe loss of MVC and accumulation ofPi – or more exactly its diprotonatedform – has been reported [19] although the exact causative natureof this relationship is unclear. WhileWilson et al [19] and Miller et al [8]have reported such a relationship in asingle group of subjects and sugges-ted that accumulation of H2PO4-could account for the failure ofmuscle force production in agree-ment with experiments conducted inskinned fibers [20], Kent-Braun et alhave recently reported that thisrelationship is modulated by age andgender, indicating that the causative



magnitude of fatigue or degree ofmetabolites accumulation. Althoughmen had a nearly twofold greaterincrease in H2PO4- during exercise,they developed no greater fatiguethan women. As a result, the slope ofthe relationship between fatigue and[H2PO4-] appears to be steeper forwomen. Similar observations werereported for pH and the overall Pi concentration. Kent-Braun et alsuggested that alterations in contrac-tile function did not explain the age-related difference in fatigue. Duringtheir moderately fatiguing exercise,whatever the group, CMAP amplitudedid not change, suggesting thatperipheral excitability was not affec-ted, whereas its duration was shorteras a sign of an increased conductionvelocity across the neuromuscularjunction or along the muscle mem-brane. Overall, neither central norperipheral (compound muscle actionpotential) played a significant role infatigue in any group. The variedmetabolic responses to exercisesuggest that the mechanisms offatigue change with age and gender[21].

III. Functional investigation ofexercising muscle using MRIAs previoulsy mentioned, the acuteeffects of exercise on muscle MRimages contrast were first reported in1988 by Fleckenstein et al. [5] andfrom then onwards, it has beenrecognized that MRI could be used todistingish between active and non-active muscle groups, thereby provi-ding some functional assessments ofexercising muscle. This informationcomes in addition to the well-knownanatomical content of T1-weightedMR images (Fig. 2).

III.A. 1H MRI Contrast differences in MR imagescome from differences in T1, T2values and nuclear density. Bone, fat,muscle, connective tissue and bloodhave different proton nuclear density,which, together with T1 and T2relaxation processes, will have asignificant impact on the final contrastobtained on the image produced. Inmuscle MR images, signals arise fromthe protons of water, lipids and bonemarrow (Fig. 3). The MR behaviour of intracellular water is believed toresult from interactions between thesurface of macromolecules and abound water layer and exchangebetween this layer and the relativelyfree cellular water. Proton motionwithin the free cellular water is itselffast enough to average out molecularinteractions; accordingly, this largefraction of cellular water may notcontribute to the MR relaxationprocess. The smaller slowly exchan-ging bound water fraction maydetermine much of the MR relaxationcharacter of muscle cells. Muscle cell

contraction involves conformationalchanges in the large contractileproteins as well as mechanical altera-tions in intracellular surfaces [27].Such surface alterations may affectthe bound water layer, which in turnmay determine MR relaxation [28]. Inaddition to alteration in the size, shapeand charge of surfaces in musclecells, contraction involves the pro-duction and translocation of ions andmetabolites. These processes wouldhave osmotic effects that alters theconcentration and the MR relaxationof water.

III.B. T2 changes in exercisingmuscleThe changes in proton signal intensi-ty cause exercised muscles to appear“lit up” in T2 weighted images asshown in figure 3. The mechanism ofthe T2 increase underlying musclefunctional MRI is still poorly under-stood, but is definitely different fromthe well-known blood oxygenationlevel dependent (BOLD) effect under-

Figure 2 Typical T1-weighted 1H MRimages recorded on a Siemens VisionPlus system (extremity coil) represen-ting transverse sections of the thigh recorded at rest in a professio-nal cyclist (A) and an age and sex-matched control subject.

Figure 3 Typical T2-weighted 1H MRimages recorded on a Siemens VisionPlus system (extremity coil) represen-ting transverse sections of the forearm recorded at rest (A) andafter a finger flexions exercise (B).Areas with a higher contrast on the image B indicate the activatedmuscles during exercise.

pH cannot change upon exercise, asclearly shown by 31P MR measure-ments [38]. In association with theabsence of intracellular acidosis, a couple of studies have reported anabsence of T2 changes in thesepatients, thereby confirming theclose relationship between MRI T2contrast changes and pH [39]. Thecause-effect relationships betweenexercise intensity, pH and MRIcontrast measurements have beenfurther illustrated from a combinedMRI and electromyography study[11]. This study has suggested thatshifts in MRI contrast after exerciseare an excellent measure of muscleuse. Indeed, EMG and T2 changeswere both different in eccentric andconcentric actions. Interestingly,both measurements were linked[11]. To go further with the determi-nation of mechanisms underlying T2 changes in activated muscles andin keeping with the hypothesis thatthese changes must result primarilyfrom altered relaxation within theactive muscle cells and not fromchanges in the extracellular fluidspace [40; 41], a series of experi-ments in humans and animals havelooked at the metabolic dependenceof T2 increase. These studies werebased on the hypothesis that anincrease in intracellular fluid duringexercise is likely to be caused by theaccumulation of osmolites such asinorganic phosphate (Pi, resultingfrom phosphocreatine (PCr) degrada-tion) and lactate. Therefore, if theosmotic expansion of an intracellularfluid compartment is linked to post-exercise T2 changes, these intensitychanges ought to vary according tothe metabolic changes. In thatrespect, it has been reported that T2changes are linked to the exerciseintensity [11] and aerobic capacity ofa given muscle [42; 43]. However,experiments conducted in marineinvertebrates have provided interes-



exercise and similar T2 changescould be expected. Fisher et al [35]demonstrated that venous occlusionhad little effect on T2 despite signi-ficant changes in muscle volume,thereby clearly indicating that exercise-induced enhanced MRI contrast doesnot result from the simple increase in fluid volume linked to increasedperfusion. More complex, probablyintracellular events would be respon-sible for exercise-induced contrastenhancement.

Exercise-induced changes in MRIT2 contrast result from some complexcombination of several processesoccurring during muscle contractionand several studies have been devotedto the determination of the mecha-nisms involved in such changes.

T2 and metabolic changes: Asindicated above, it has been knownfrom invasive studies that the volumeof exercising muscle increases as aresult of redistribution of body water[30; 34]. While changes in musclevolume as a factor involved in post-exercise T2 changes have beenrefuted on the basis of venous occlu-sion experiments [11], it has beenclearly demonstrated that T2 changeswere graded with exercise intensityduring dorsiflexion exercises [35],thereby indicating that T2 changesduring exercise are dependent ongenerated force. Furthermore, adirect relationship between changesin osmotically active metabolites –such as lactate and inorganic phos-phate – have also been correlatedwith the extent of T2 changes [36;37], indicating that they are likely tobe related to fluid-shift osmoticallydriven. This relationship between theextent of T2 and pH changes inexercising muscle has been confirmedfrom a functionnal analysis of McArdlepatients. Such patients suffer fromglycogenolysis deficiency and areunable to produce lactate [38]. As aresult of this deficiency, intracellular

lying brain functional MRI investiga-tions. T2 measurements in musclehave been performed in order toelucidate the signal shift in musclethat is related to exercise. Moreparticularly, MRI measurementsfocuse on the intrinsic property ofwater and its exchange betweencompartments as well as the bindingcapacity of the water molecule tosubcellular structures. Changes in T2accounting for the discriminationbetween active and inactive musclesresult from the dependence ofrelaxation on the local molecularenvironment of the nuclei understudy [28]. Given that T2 changesaffect exclusively muscle and neitherfat nor bone marrow, it is reasonableto conclude that the signal changesarise from one or more of the watercompartments in muscle i.e. fromintra and extracellular spaces.

T2 and volume changes: As anexample, different T2 values inresting muscle have been reported asan indication of different size inextracellular spaces [29]. Increasedperfusion would therefore seem anobvious factor, which could accountfor exercise-induced T2 changes. The volume of exercising muscle isknown to increase because of redis-tribution of body water [30; 31]. Thisredistribution could be attributed toincreased perfusion and the produc-tion and translocation of ionic specieswhich could alter the osmotic beha-viour of muscle cells [32; 33]. Lowintensity exercise is believed to beassociated with an increase in extra-cellular water, whereas high intensityexercise is primarily associated withchange in intracellular water [31;34]. Changes in either intra or extra-cellular water would be expected toalter the relaxation characteristics ofthe excited nuclei in a muscle sample[28]. However, changes in musclevolume can be induced by venousocclusion in the same extent as



ting information as to the relationbetween T2 and metabolic changes[44]. Contraction of tail muscle ofcrayfish (an osmoregulator specieswith osmolarity near 340mosM) waslinked to T2 changes, whereascontraction of lobster tail (a marineosmoconformer with osmolarityequal to that of seawater i.e. 1 osmolar)was not linked to any. These resultsclearly point to T2 changes beinglinked to a redistribution of tissuefluid caused by accumulation ofintracellular osmolites as previoulsyindicated in rat muscles [43]. Howe-ver, pH changes occurred only inlobster tail muscle upon contractionand not in crayfish tail muscle,thereby suggesting that muscleacidification per se is not a necessaryor dominant cause of the T2 increaseduring stimulation [44]. Overall, onecan conclude that the T2 increase inactive mamalian muscles is caused byosmotically driven fluid shifts bet-ween subcellular compartments.However, in order to exploit thesechanges as either a mapping tool oran index of exercise intensity, onehas to address the issue of normali-sing the T2 changes between diffe-rent muscles and different individu-als. Indeed, heterogeneous exercise-induced T2 changes have beenreported among subjects with diffe-rent training status or among muscleswithin the same subject with no clearinterpretation regarding the under-lying mechanisms [36; 45]. Forinstance, we have observed a largerange of T2 changes after exercise inprofessional cyclists with similaraerobic aptitudes (Fig. 4). It has beenclearly shown from comparativeanalysis between eccentric andconcentric exercises that the T2increase after exercise is not depen-dent on absolute work rate per se butis linearly linked to exercise intensityrelative to maximum aerobic power[42; 46]. In addition, Prior et al [43]

have recently reported that the T2increase after exercise varied inverse-ly with the known aerobic aptitudes.In keeping with the results publishedby Reid et al [42], such a standardisa-tion should allow reliable compari-son.

IV. Conclusion

For many years, there has beenconsiderable interest in trying todetermine whether or not biochemi-cal changes occurring in exercisingmuscle actually contributed to thesubjective manifestation of fatigue.The development of fatigue is attri-butable to both central and periphe-ral factors. The relative contributionsof these factors may be estimatedusing a combination of voluntary andelectrically stimulated force measure-ments, 31P MRS and surface EMG. This type of approach should impro-ve and shed some light on our under-standing of the mechanisms ofhuman muscle fatigue by simultane-ously assessing functions at thevarious stages along the pathway offorce production.

With regard to the mapping of mus-cles activated during exercise, inve-stigation of T2 changes using MRIprovides reliable information as longas standardisation procedure areproperly used. Although it has beenquite clearly established that themechanisms underlying these T2changes are linked to altered relaxa-tion within the active muscle cellslikely due to accumulation of osmoticmetabolites, the exact nature ofthese metabolites still remain to beprecisely identified.

References

[ 1 ] Hoult, D. I.; Busby, S. J.; Gadian, D. G.;Radda, G. K.; Richards, R. E.; Seeley, P. J. (1974).Observation of tissue metabolites using 31Pnuclear magnetic resonance. Nature 252, 285-7.

[ 2 ] Argov, Z.; Arnold, D. L. (2000). MR spectroscopy and imaging in metabolicmyopathies. Neurol Clin 18, 35-52.

[ 3 ] Bendahan, D.; Mattei, J. P.; Kozak-Ribbens, G.; Cozzone, P. J. (2002). [Non-invasive investigation of muscle diseasesusing 31P magnetic resonance spectroscopy:potential in clinical applications]. Rev Neurol (Paris) 158, 527-40.

[ 4 ] Cozzone, P. J.; Vion-Dury, J.; Bendahan, D.; Confort-Gouny, S. (1996).[Future path of magnetic resonance spectroscopy in clinical medicine]. Rev. Prat. 46, 853-8.

[ 5 ] Fleckenstein, J. L.; Canby, R. C.; Parkey, R. W.;Peshock, R. M. (1988). Acute effects of exerciseon MR imaging of skeletal muscle in normalvolunteers. AJR Am J Roentgenol 151, 231-7.

[ 6 ] Bigland-Ritchie, B.; Jones, D. A.; Hosking, G. P.; Edwards, R. H. (1978). Central and peripheral fatigue in sustainedmaximum voluntary contractions of humanquadriceps muscle. Clin. Sci. Mol. Med. 54, 609-14.

[ 7 ] Degroot, M.; Massie, B. M.; Boska, M.;Gober, J.; Miller, R. G.; Weiner, M. W. (1993).Dissociation of [H+] from fatigue in humanmuscle detected by high time resolution 31P-NMR. Muscle Nerve 16, 91-8.

[ 8 ] Miller, R. G.; Giannini, D.; Milner-Brown,H. S.; Layzer, R. B.; Koretsky, A. P.; Hooper, D., et al. (1987). Effects of fatiguing exercise on high-energy phosphates, force, and EMG:evidence for three phases of recovery. Muscle Nerve 10, 810-21.

Figure 4 Relative T2 changes (expressed in % of the value measu-red at rest) in exercising muscles (VL:vastus lateralis, RF : rectus femoris,VM: vastus medialis) of six professio-nal cyclists. It is noteworthy thatdespite their similar aerobic aptitudes,T2 changes largely vary from onesubject to another.

[ 36 ] Weidman, E. R.; Charles, H. C.; Negro-Vilar, R.; Sullivan, M. J.; MacFall, J. R.(1991). Muscle activity localization with 31Pspectroscopy and calculated T2-weighted 1H images.Invest Radiol 26, 309-16.

[ 37 ] de Kerviler, E.; Leroy-Willig, A.; Jehenson, P.; Duboc, D.; Eymard, B.; Syrota, A.(1991). Exercise-induced muscle modifications:study of healthy subjects and patients withmetabolic myopathies with MR imaging and P-31 spectroscopy. Radiology 181, 259-64.

[ 38 ] Bendahan, D.; Confort-Gouny, S.; Kozak-Ribbens, G.; Cozzone, P. J. (1992). 31-P NMR characterization of the metabolicanomalies associated with the lack of glycogenphosphorylase activity in human forearmmuscle. Biochem Biophys Res Commun 185, 16-21.

[ 39 ] Fleckenstein, J. L.; Haller, R. G.; Lewis, S. F.; Archer, B. T.; Barker, B. R.; Payne, J.,et al. (1991). Absence of exercise-induced MRI enhancementof skeletal muscle in McArdle's disease. J Appl Physiol 71, 961-9.

[ 40 ] Conley, M. S.; Foley, J. M.; Ploutz-Snyder, L. L.; Meyer, R. A.; Dudley, G. A.(1996). Effect of acute head-down tilt onskeletal muscle cross-sectional area and protontransverse relaxation time. J Appl Physiol 81, 1572-7.

[ 41 ] Ploutz-Snyder, L. L.; Nyren, S.; Cooper, T. G.;Potchen, E. J.; Meyer, R. A. (1997). Different effects of exercise and edema on T2relaxation in skeletal muscle. Magn Reson Med 37, 676-82.

[ 42 ] Reid, R. W.; Foley, J. M.; Jayaraman, R. C.;Prior, B. M.; Meyer, R. A. (2001). Effect of aerobic capacity on the T(2) increasein exercised skeletal muscle. J Appl Physiol 90, 897-902.

[ 43 ] Prior, B. M.; Ploutz-Snyder, L. L.; Cooper, T. G.; Meyer, R. A. (2001). Fiber type and metabolic dependence of T2increases in stimulated rat muscles. J Appl Physiol 90, 615-23.

[ 44 ] Meyer, R. A.; Prior, B. M.; Siles, R. I.;Wiseman, R. W. (2001). Contraction increases the T(2) of muscle infresh water but not in marine invertebrates. NMR Biomed 14, 199-203.

[ 45 ] Le Rumeur, E.; Carre, F.; Bernard, A. M.;Bansard, J. Y.; Rochcongar, P.; De Certaines, J. D.(1994). Multiparametric classification of muscleT1 and T2 relaxation times determined bymagnetic resonance imaging. The effects ofdynamic exercise in trained and untrainedsubjects. Br J Radiol 67, 150-6.

[ 46 ] Shellock, F. G.; Fukunaga, T.; Mink, J. H.;Edgerton, V. R. (1991). Acute effects of exercise on MR imaging ofskeletal muscle: concentric vs eccentric actions. AJR Am J Roentgenol 156, 765-8.



[ 22 ] Vestergaard-Poulsen, P.; Thomsen, C.;Sinkjaer, T.; Henriksen, O. (1994). Simultaneous 31P NMR spectroscopy and EMGin exercising and recovering human skeletalmuscle: technical aspects. Magn. Reson. Med. 31, 93-102.

[ 23 ] Laurent, D.; Portero, P.; Goubel, F.;Rossi, A.(1993). Electromyogram spectrum changesduring sustained contraction related to protonand diprotonated inorganic phosphate accumu-lation: a 31P nuclear magnetic resonance studyon human calf muscles. Eur. J. Appl. Physiol. Occup. Physiol. 66, 263-8.

[ 24 ] Lindstrom, L.; Kadefors, R.; Petersen, I. (1977). An electromyographic indexfor localized muscle fatigue. J. Appl. Physiol. 43, 750-4.

[ 25 ] Kent-Braun, J. (1999). Central and peripheral contributions to musclefatigue in humans during sustained maximaleffort. Eur. J. Appl. Physiol. 80, 57-63.

[ 26 ] Cady, E. B.; Jones, D. A.; Lynn, J.;Newham, D. J. (1989). Changes in force and intracellular metabolitesduring fatigue of human skeletal muscle. J. Physiol. (Lond.) 418, 311-25.

[ 27 ] Fung, B. M. (1977). Correlation of relaxation time with watercontent in muscle and brain tissues. Biochim Biophys Acta 497, 317-22.

[ 28 ] Fullerton, G. D.; Potter, J. L.; Dornbluth, N. C. (1982). NMR relaxation ofprotons in tissues and other macromolecularwater solutions. Magn Reson Imaging 1, 209-26.

[ 29 ] Polak, J. F.; Jolesz, F. A.; Adams, D. F.(1988). Magnetic resonance imaging ofskeletal muscle. Prolongation of T1 and T2subsequent to denervation. Invest Radiol 23, 365-9.

[ 30 ] Lundvall, J.; Mellander, S.; Westling, H.;White, T. (1972). Fluid transfer between bloodand tissues during exercise. Acta Physiol Scand 85, 258-69.

[ 31 ] Sjogaard, G.; Saltin, B. (1982). Extra- and intracellular water spaces in musclesof man at rest and with dynamic exercise. Am J Physiol 243, R271-80.

[ 32 ] Rowell, L. B. (1988). Muscle blood flow in humans: how high can it go?Med Sci Sports Exerc 20, S97-103.

[ 33 ] altin, B.; Sjogaard, G.; Gaffney, F. A.;Rowell, L. B. (1981). Potassium, lactate, and water fluxes in human quadriceps muscleduring static contractions. Circ Res 48, I18-24.

[ 34 ] Sjogaard, G. (1986). Water and electrolyte fluxes during exercise and theirrelation to muscle fatigue. Acta Physiol Scand Suppl 556, 129-36.

[ 35 ] Fisher, M. J.; Meyer, R. A.; Adams, G. R.;Foley, J. M.; Potchen, E. J. (1990). Direct relationship between proton T2 andexercise intensity in skeletal muscle MR images. Invest Radiol 25, 480-5.

[ 9 ] Moritani, T.; Muro, M.; Kijima, A. (1985).Electromechanical changes during electricallyinduced and maximal voluntary contractions:electrophysiologic responses of differentmuscle fiber types during stimulated contrac-tions. Exp. Physiol. 88, 471-483.

[ 10 ] Bouissou, P.; Estrade, P. Y.; Goubel, F.;Guezennec, C. Y.; Serrurier, B. (1989). Surface EMG power spectrum and intramus-cular pH in human vastus lateralis muscleduring dynamic exercise. J. Appl. Physiol. 67, 1245-9.

[ 11 ] Adams, G. R.; Duvoisin, M. R.; Dudley, G. A.(1992). Magnetic resonance imaging and electromyography as indexes of musclefunction. J Appl Physiol 73, 1578-83.

[ 12 ] Hagberg, M.; Ericson, B. E. (1982).Myoelectric power spectrum dependence onmuscular contraction level of elbow flexors.Eur. J. Appl. Physiol. Occup. Physiol. 48, 147-56.

[ 13 ] Moritani, T.; Muro, M.; Nagata, A. (1986).Intramuscular and surface electromyogramchanges during muscle fatigue. J. Appl. Physiol. 60, 1179-85.

[ 14 ] Horita, T.; Ishiko, T. (1987). Relationships between muscle lactate accumulation and surface EMG activities duringisokinetic contractions in man. Eur. J. Appl. Physiol. Occup. Physiol. 56, 18-23.

[ 15 ] Mortimer, J. T.; Magnusson, R.; Petersen, I.(1970). Conduction velocity in ischemic muscle:effect on EMG frequency spectrum. Am. J. Physiol. 219, 1324-9.

[ 16 ] Moussavi, R. S.; Carson, P. J.; Boska, M.D.; Weiner, M. W.; Miller, R. G. (1989). Nonmetabolic fatigue in exercising humanmuscle. Neurology 39, 1222-6.

[ 17 ] Bendahan, D.; Jammes, Y.; Salvan, A. M.;Badier, M.; Confort-Gouny, S.; Guillot, C., et al.(1996). Combined electromyography-31P-magnetic resonance spectroscopy study ofhuman muscle fatigue during static contrac-tion. Muscle Nerve 19, 715-21.

[ 18 ] Bendahan, D.; Badier, M.; Jammes, Y.;Confort-Gouny, S.; Salvan, A. M.; Guillot, C., et al. (1998). Metabolic and myoelectricaleffects of acute hypoxaemia during isometriccontraction of forearm muscles in humans: a combined 31P- magnetic resonance spectroscopy-surface electromyogram (MRS-SEMG) study. Clin Sci (Colch) 94, 279-86.

[ 19 ] Wilson, J. R.; McCully, K. K.; Mancini, D. M.;Boden, B.; Chance, B. (1988). Relationship of muscular fatigue to pH anddiprotonated Pi in humans: a 31P-NMR study. J. Appl. Physiol. 64, 2333-9.

[ 20 ] Nosek, T. M.; Fender, K. Y.; Godt, R. E.(1987). It is diprotonated inorganic phosphatethat depresses force in skinned skeletal musclefibers. Science 236, 191-3.

[ 21 ] Kent-Braun, J. A.; Ng, A. V.; Doyle, J. W.;Towse, T. F. (2002). Human skeletal muscleresponses vary with age and gender duringfatigue due to incremental isometric exercise. J. Appl. Physiol. 93, 1813-23.

Licenses:

Is prostate MRS a separatepackage?Prostate spectroscopy is a part of the 1H CSI-option and is includedwith the syngo MR 2004A/2004V.

Multinuclear The multinuclear option contains thehardware and system driving, tune-up and adjustment SW for obtainingimages and spectra of the followingnuclei: 13C, 19F, 3He, 7Li, 23Na, 39K, 17O (3T only), 129XeAdditionally, proton NOE and pulsedCW proton decoupling experimentscan be run.

The multinuclear option is avail-able for Symphony and Sonatasystems (including upgrades) fromsyngo MR 2002B onwards, and forTrio and Allegra with 2004A. For theAvanto system, it will be availablewith the next major SW release.

The multinuclear option itself isintended for research users who are willing to create dedicated setupsusing special sequences and coils.Recommended multinuclear coilproviders for Siemens systems are: • Advanced Imaging Research,

Cleveland, Ohio, USA.http://www.advimg.com

• Bruker Biospin MRI, Ettlingen,Germany. http://www.bruker-biospin.de

• Rapid Biomedical, Wuerzburg,Germany. http://www.rapidbiomed.de/

The application development environment IDEA is required forcreating special multinuclear sequences.


CLINICALMR SPECTROSCOPY

Marianne Vorbuchner

MRS Application DevelopmentSiemens Medical SolutionsErlangen, Germany

FAQs

For prostate spectroscopy, should we use a -2.1 delta shift toplace Citrate in the center?Yes, the Citrate peak (= 2.6 ppm) can be used: set the delta shift at -2.1(4.7 minus 2.1 = 2.6 ppm). The rangebetween Citrate and Choline can also be used, in this case set the deltafrequency to -1.8.

Is there a separate parameter for a delta shift for the sat bands,according to what signal we wantto eliminate?The regional sat bands in the prostateprotocol are used to suppress thelocal fat signal (set at -3.4) and NOTthe water signal. Use a minimal sat band thickness in order to avoidchemical shift artifacts, and keep in mind that the first sat band is themost effective (this is time depen-dent).

Do the 4 invisible sats apply only to CSI, or also to svs? Are these 4 sats automatically on or is there a key that has to be toggled on?This new feature is the “fully excitedVOI” (shown also on the ContrastCard). If you select the check boxthere are 4 internal, invisible satura-tion bands: this enables the acquisi-tion of a bigger volume in order tokeep the chemical shift artifactbetween water/fat a little lower. Fullyexcited VOI is only available for CSIspin echo protocols.

Spectral suppression pulses, what are they for?Using the CSI_SE sequence with the‘spectral suppression’ feature you caneither suppress lipid signals or watersignals or both. These pulses are veryeffective global suppression pulses,mainly used for prostate applications.Note, however, that the minimumecho time is only 90 ms.

Matrix Spectroscopy:

When is the matrix spectroscopy available?

Matrix spectroscopy (signal combina-tion of MRS data originating frommulti-array coils) is available with theTim-SW syngo MR 2004V.

All localization methods – csi, svs and fid – are supported.

A stripped-down version of thisproduct feature is available as a WIPpackage for the software versions2002B, 2003T and 2004A (MultipleArray).

A transmit/receive head coil is available on 1.5 Tesla MAGNETOM Symphony/Sonata and on theMAGNETOM Avanto system.

Delta Frequency / OVS:

For brain spectroscopy, should we use a -2.7 delta shift to placeNAA (2.0 ppm) in the center?This depends on the pathology. For example, for a lateral lesion closeto the bone, the delta frequency can shift to fat (4.7 minus 3.4 = 1.3ppm).


CLINICALMR SPECTROSCOPY

31P MRS31P MRS is a ready-to-use applicationpackage for heart, liver and calfmuscle 31P-spectroscopy. It containsthe double resonant 31P / 1H heart-liversurface coil, and dedicated acquisi-tion and post-processing protocolsbased on the csi_fid and the fidsequence. 1H NOE and decouplingparameters are activated or availablein those protocols.

The 31P MRS option requires themultinuclear option among otheroptions. It is available for Symphonyand Sonata systems (includingupgrades) from syngo MR 2002Bonwards and will be available for theAvanto system with the next majorSW release. It is planned to offer the 31P MRS option also with our 3Tsystems Trio and Allegra.

Shim:

How is the best shim valuecalculated during interactiveshimming? The best shim value is calculated as a maximum of T2*. The T2* is morestable than the Full Width Half Maxi-mum. In the optimal case, FWMHshould be as small as possible, andT2* as large as possible.

How can the shim result be controlled?To achieve a highest possible spectraquality, a homogeneous magneticfield is required. Therefore “AdvancedShim” is the default setting in allspectroscopy protocols. The AdvancedShim field Map measurement is aniterative procedure. On all 1T, 1.5Tand 3T systems three different shimmodes are available: push-button

mode, semi automatic mode, manualmode (the Advanced Shim is inde-pendent from the hardware option“Advanced High OrderShim”).

Within this shim modes, there isthe helpful possibility of using thesemi automatic shim. In this mode,all adjustments are automaticallyexecuted. Then the user can controland improve the results if necessarybefore the real start of the measure-ment. This is recommended in criticalanatomic areas, or in off-centerpositions.

Water Suppression:

Please distinguish between watersuppression during measurementand the post-processing step ofwater reference processing.

Sequence:

What is the difference between our spin echo sequence and thePRESS technique? The spin echo and the PRESStechnique are precisely the samething.

Measurement:

Please explain how the new watersuppression adjustment of syngoMR 2004A works?The water suppression in syngo MR2002B was iterative. In the newversion, this is replaced by non-iterative searching for the optimumflip-angle. The aim is to shorten themeasurement time and to increasethe robustness of the adjustmentprocedure.

Export Rawdata:

How do we get data – such asspectra and even raw data – fromthe MRI system?You can copy the whole patient/study/series, including spectra imageor raw data, from the Patient Browserto a CD-Rom. Export to offline in a DICOM or BITMAP format is alsopossible via Patient Browser. Fromthe Spectroscopy Card, the Export tooffline function allows the raw data-export in a private ASCII format(explained in the Spectro ApplicationGuide, VA21B).Important: The switch “implicit rawdata transfer” in the service UI must be switched on (Local service / Configuration / Next / Dicom / General / > ).

Postprocessing:

In syngo MR 2002B boundary CSIvoxels may not be evaluated, and themessage “water suppression failed” isdisplayed.

The problem will be solved in SW2004A. For older SW versions, openthe post-processing protocol andswitch off the first step “water refe-rence processing”. The cause of thisproblem is a strong water peak orwater hump. The software will searchthe highest signal at 4.7 +/- 0.5 ppm.If there is a “bad” water signal, thenthe error message appears.

continue to demonstrate our com-mitment to this tradition by offeringsimple, easy and smart upgrades toserve almost any MAGNETOM productthat you may be using. We are proud to offer the following Elevateprograms:

TECHNOLOGY Elevate. Such as Tim*,for our MAGNETOM Symphonyand Trio systems will increase yourperformance with the latest in MRadvancements. Your Symphony onlyneeds the Quantum gradient packagein order to be able to move into theTim world of performance. Pleasecontact your local sales representa-tive regarding the Quantum gradientupgrade and the additional benefitsyou will experience.

SYSTEM Elevate for our classicproducts such as the MAGNETOMVision and Impact will utilize yourexisting magnet vessel technologyhowever transformed to the latestIPA and syngo performance levels.Restoring Life and protecting yourinvestment. Elevate to a MAGNETOMHarmony or Symphony or Sonata.

MANAGED Elevate simplifies the replacement of your previousgeneration system and provides the latest cutting-edge technologyand performance. Elevate to a MAGNETOM Avanto. The first Tim system.

There has never been a better time to consider upgrading your MAGNETOM system.

Please contact your local salesorganization and inquire as to which MAGENTOM Elevate programwill best meet your needs.


LIFECUSTOMER CARE

Sean HarrisonInstalled Base Manager

MR Marketing,Siemens AG Medical Solutions,Erlangen, Germany

Life

We would like to introduce a keyingredient of Life with respect to MR.MAGNETOM Elevate. Our upgradeprogram.

There are currently many challen-ges confronting modern diagnosticimaging. Declining reimbursements,staffing shortages and decreasedcapital budgets are among theseexternal factors which affectadministrative decisions regardingthe purchase of imaging systems.These market conditions supportexploring cost effective alternativeswhich maximize system performanceand workflow optimization.

MAGNETOM Elevate. It’s all aboutLife. And Life is about getting themost out of your MR investment andmaintaining cutting-edge technologyat all levels. Siemens has a provenhistory of MR upgrades spanningover 25 years! Beginning with thefirst MRI systems installed in the early80s, we have consistently offered MR upgrades to support EACH plat-form since. It’s a tradition in MR. We

Life. Life is our customer care solution.There are seven key programs to Life.In the most general sense our pro-grams support both the continuousdevelopment of your skills, as well asthe continuous development of yourproduct. Both essential to helpingyou extract every bit of value fromyour investment. These seven programsare focused on helping you meetyour clinical and business objectives.Delivering Proven Outcomes. Today.And far into the future.


Life

Get the most from your investment

Continous development with Life

Continous development of your skills

Continous development of your productivity

Continous development of your technology


LIFECUSTOMER CARE

MAGNETOM Flash – Reader Service

Letters to the Editor – We welcome your comments

about the content of MAGNETOM Flash. Please send

comments to the Editor. Include your name, address,

and phone number or e-mail.

World Wide Web – Visit us at

www.SiemensMedical.com and

www.SiemensMedical.com/MAGNETOM-World

These sites provide information about all Siemens

medical products.

Publish articles? – You are invited to publish articles in

the newsletter to share your experience with

MAGNETOM MR users all over the world. To submit an

article please contact the Editor.

Subscription – You have seen the newsletter and want

to get it on a regular basis? In the US, please contact the

Applications Helpline (phone 800-888-SIEM) and give

us your name and business address (no home

addresses, please). Outside the US, MAGNETOM Flash is

distributed through the local Siemens offices. Please

contact the Editor and we will make sure that you are

included on your local support office’s distribution list.

Editor

Ali Nejat Bengi, M.D., [email protected]

Published by

Siemens AG Medical Solutions

P.O. Box 3260, D-91052 Erlangen

Correspondence and

International Distribution

Ali Nejat Bengi, M.D., Editor in Chief

MAGNETOM FLASH

Siemens AG Medical Solutions

MR Marketing

Karl-Schall-Str. 6

D-91052 Erlangen

Phone: 49 - 91 31 - 84 - 75 99

Fax: 49 - 91 31 - 84 - 21 86

US Distribution

MR APPLICATIONS HELPLINE

Siemens Uptime Service Center

110 MacAlyson Court

Cary, NC 27511

Phone: 800 - 888 - SIEM

Fax: 919 - 319 - 28 64

All articles represent the techniques and opinions

of the authors and may not represent specific

recommendations or endorsements from Siemens

Medical Solutions. Contact the authors directly for

further information about their techniques and

opinion.

Siemens reserves the right to modify the design and

specifications contained herein without prior notice.

Please contact your local Siemens sales representative

for the most current information.

Original images always lose a certain amount of detail

when reproduced.

This brochure refers to both standard and optional

features. Availability and packaging of options varies by

country and is subject to change without notice.

Some of the features described are not available for

commercial distribution in the US.

The information in this document contains general

descriptions of the technical options available, which

do not always have to be present in individual cases.

The required features should therefore be specified in

each individual case at the time of closing the contract.

Please contact in the USA:Siemens Medical Solutions USA, Inc.51 Valley Stream Parkway Malvern, PA 19355Tel.: +1 888-826-9702Tel.: 610-448-4500Fax: 610-448-2254

in Asia:The Siemens Centre 60 MacPherson Road Singapore 348615 Tel.: +65 6341 0990 Fax: +65 6778 6722

in Japan:Siemens-AsahiMedical Technologies Ltd.Takanawa Park Tower 14 F20-14, Higashi-Gotanda 3-chomeShinagawa-kuTokyo 141-8641(03) 54 23 40 01

Or contact your local SiemensSales Representative

Siemens AG, Medical SolutionsHenkestr. 127, D-91052 ErlangenGermanyTelephone: +49 9131 84-0www.siemens.com/medical

Siemens AG, Medical SolutionsMagnetic ResonanceHenkestr. 127, D-91052 Erlangen, GermanyTelephone: ++49 9131 84-0www.siemens.com/medical

© 2004 Siemens Medical Solutions

Order No. A91100-M2220-F691-7-7600

Printed in Germany

GP 65558 WS 050420.

[28] magnetom flash_28_jan 2004

Documents