Allen D. Elster, MD #{149}James C. King, MD #{149}Vincent P. Mathews, MD #{149}Craig A. Hamilton, PhD

Cranial Tissues: Appearance atGadolinium-enhanced and Nonenhanced MRImaging with Magnetization Transfer Contrast’

Neuroradiology

PURPOSE: To determine the relativecontrast of normal cranial tissues at

magnetization transfer (MT) spin-echo magnetic resonance (MR) imag-ing.

MATERIALS AND METHODS: MR

imaging at 1.5 T was performed withconventional spin-echo techniqueswithout and with off-resonance MTsaturation pulses. The signal intensi-ties of normal cranial tissues weremeasured in 10 healthy volunteers onspin-density- and T2-.weighted im-ages and in 10 patients on Ti-weighted images obtained before andafter administration of gadopentetatedimeglumine.

RESULTS: MT saturation produced asignificant (P < .Oi) reduction in sig-nal from all tissues except cerebrospi-nal fluid and fat. Several gray matterstructures had higher signal intensitythan white matter on Ti-weightedMT images. After administration ofgadopentetate dimeglumine, imagingwith the MT sequence increased vi-sualization of normally enhancingstructures.

CONCLUSION: MT saturationpulses produce new patterns of tissuecontrast that differ substantially from

those seen on conventional spin-echoimages.

Index terms: Brain, MR. 10.121417 ‘ Magnetic

resonance (MR), magnetization transfer contrast

Radiology 1994; 190:541-546

I From the Department of Radiology, Bow-man Gray School of Medicine, Wake Forest Uni-versity, Medical Center Blvd, Winston-Salem,

NC 27157-1088. From the 1993 RSNA scientificassembly. Received August 10, 1993; revisionrequested September 24; revision received Octo-ber 4; accepted October 8. Address reprint re-quests to A.D.E.

C RSNA, 1994See also the article by Mathews et al (pp 547-

552) in this issue.

M AGNETIZATION transfer (MT) is arelatively new magnetic reso-

nance (MR) imaging technique thatallows suppression of signal from tis-sues in which macromolecular inter-actions with water molecules contrib-ute substantially to relaxation (1-3).Although many variations of thismethod are possible, the usual MTtechnique consists of applying off-resonance radio-frequency saturationpulses selectively to saturate the pool

of macromolecular protons (4,5). MTsaturation of this macromolecularpool then modulates the MR signalfrom the nearby water molecules, pre-sumably by means of dipolar cross-coupling and chemical-exchange in-teractions (6-8).

Findings in preliminary studies

have demonstrated that the MT tech-nique may be useful in several cranialimaging applications: for improvingthe visualization of small vessels attime-of-flight MR angiography (9-11);for increasing tissue contrast in gradi-ent-echo imaging (12); for demon-strating the internal structure of theeye (13); for assessing the integrity ofmyelin (14-16); for detecting cerebralinfarction (17); and for delineating avariety of intracranial mass lesions,

with and without gadolinium en-hancement (4,18-22). As MT tech-niques become more widely dissemi-nated and used for general diagnosticpurposes, it becomes increasingly im-portant to recognize and documenthow these pulses alter the image con-trast of normal intracranial tissues.

MATERIALS AND METHODS

MR imaging was performed on a 1.5-1unit (Signa; GE Medical Systems, Milwau-kee, Wis). An MT pulse was produced bymodifying the standard chemical-shift se-lective saturation pulse available on thissystem (Fig 1). The resultant MT satura-tion pulse had a bandwidth of 220 Hz andan envelope composed of a 16-msec, apo-dized smc function with two side lobes.

The frequency offset of this pulse was setat 1,200 Hz downfield from the water reso-nance. The pulse amplitude was maxi-mized to produce a peak radio-frequencyfield flux density (B1) of 7.3 p.1. After each

MT pulse, gradient homospoiling was per-formed by turning on the y-axis (phase-encoding) gradient to maximum ampli-tude for about 5 msec. (The purpose ofthis spoiling pulse was to prevent possible

interference patterns with the next radio-frequency pulse.) Allowing for gradientramp times, the interval between the endof the MT pulse and the beginning of the900 pulse was approximately 6 msec.

The MT pulse sequence was tested in 20subjects: (a) 10 healthy volunteers (fivemen and five women, aged 24-27 years)who underwent noncontrast imaging onlyand (b) 10 patients (three men and sevenwomen, aged 31-76 years) who under-went gadolinium-enhanced imaging aspart of their work-up for suspected neuro-logic disease. The conventional precon-trast MR images of these 10 patients werejudged to be unequivocally normal bythree experienced neuroradiologists(A.D.E., J.C.K., V.P.M.). Patients were en-rolled for the contrast material-enhancedportion of the study so that volunteerswould not be exposed to any possible risk,however small, of an adverse reaction togadolinium. Overall, therefore, our cohortof 20 subjects comprised eight men and 12

women, aged 27-76 years (median, 38’/2

years). Informed consent for the adminis-tration of gadolinium and/or for the test-ing of the MT sequences was obtainedfrom each subject. Approval was obtainedfrom our local institutional review boardand from GE Medical Systems to operatethe MR imager in an investigational mode.Specific absorption rates were carefullymonitored in every case and remainedbelow limits mandated by the U.S. Foodand Drug Administration (average, 3.2W/kg for the whole head and 8.0 W/kgfor any single gram of cranial tissue).

Spin-density- and 12-weighted spin-

echo imaging were performed with andwithout MT pulses in the 10 volunteers,with use of the following parameters:plane of imaging, transaxial; repetition

Abbreviation: MT = magnetization transfer.

‘4-’

&� ,

.�#{149}-,-�4

‘ 1’

a.

4�

‘��:1�N�

.:

b.

,� �

d/s �

/A.-

�

�:T’ ‘K’ �S\

5-’-,’ � ,�

\�. � S �

.:�::a!# ssK�

!%�: � $#{149} “p

\�4

C. �%�L.I:,%,�I�v d.

Figure 2. MR images of 27-year-old femal1� volunteer. (a) On spin-density-weighted image

obtained without MT. cerebrospinal fluid has relatively low signal intensity and fat has rela-tively high signal intensity. (b) On spin-density-weighted image obtained with MT, both cere-

brospinal fluid and scalp fat have relatively higher signal intensity compared with brain. Thewhite matter (MT ratio = 28%) is also suppressed somewhat more than the deep gray matter

nuclei (MT ratios = 20%-23%). (c) On T2-weighted image obtained without MT, cerebrospinalfluid has relatively high signal intensity and scalp fat has intermediate signal intensity. (d) On

T2-weighted image obtained with MT, the brain parenchyma has lower signal intensity, mak-

ing both cerebrospinal fluid and fat appear relatively brighter.

I542 #{149}Radiology February 1994

MT 90’ 180’

RF

Section

Read �

Phase homospoil

Figure 1. Diagram of the pulse timing ofthe MT spin-echo sequence. RF = radio fre-

quency.

time, 3,000 msec; echo times, 30 and 90

msec; field of view, 22 cm; imaging matrix,

192 x 256; 20 sections with thickness of 5

mm; intersection gap, 2 mm; and one sig-

nal averaged, with partial phase-conjugatesymmetry. The total imaging time for both

conventional spin-echo and MT-saturatedlong repetition time sequences was 7 mm-utes 48 seconds. Receive and transmit at-

tenuator settings were maintained at con-stant levels within each pair of comparable

pulse sequences for each subject.In the 10 patients, Ti-weighted images

were obtained both with and without MTsaturation pulses, with use of the follow-ing parameters: plane of imaging, trans-

axial; repetition time, 600 msec; echo time,15 msec; field of view, 22 cm; imaging ma-

trix, 256 x 192; 20 sections with thickness

of 5 mm; intersection gap, 2 mm; and onesignal averaged. To obtain 20 sections with

the MT pulse, two separate 10-section ac-quisitions were performed, each with a

35% MT duty cycle (ie, the fraction of time

the MT pulse was activated during eachrepetition time interval). The imaging timefor each conventional Ti-weighted spin-echo study was therefore 2 minutes 4 sec-onds, whereas each MT-saturated study

required 4 minutes 7 seconds. Gadopen-tetate dimeglumine (Magnevist; BerlexLaboratories, Wayne, NJ) was adminis-tered by slow intravenous infusion to eachpatient at a dose of 0.i mmol/kg. After adelay of approximately 5 minutes, Ti-weighted images were obtained with andwithout MT saturation; in five patients,Ti-weighted images with MT were ob-tamed first, and in the remaining five pa-tients Ti-weighted images without MTsaturation were obtained first.

The images with each paired sequencewere obtained with identical window andlevel settings and were visually assessedby the same three experienced neuroradi-

ologists in side-by-side, nonblinded corn-

parisons. To minimize the biases inherent

in any such visually based evaluation, sixanatomic regions of interest were selectedfor quantitative signal analysis: cerebrospi-

nal fluid, scalp fat, cerebral white matter,caudate, and putamen (for the long repeti-tion time images), with the addition of the

pituitary gland (for the Ti-weighted im-

ages only). These measurements were allobtained by a single investigator (J.C.K.),who, with use of conventional imager

software, placed an oval cursor over each

designated region of interest and obtained

mean signal intensity measurements. Thesize and location of this region-of-interest

cursor was not changed between each set

of paired acquisitions; visual assessment ofthe images confirmed that no noticeablepatient motion or image misregistration

had occurred. For all quantitative determi-nations of signal intensity, the median of

at least three independently obtained

measurements was used.For each tissue and pulse sequence, an

MT ratio was calculated with the formula

MTR = (SI0 - SI,,,)! SIC, where MTR is MT

ratio, SI� is the signal intensity of the tissue

before the MT pulse has been applied, and

51� is the signal intensity after the MT

pulse has been applied. Calculation of the

MT ratio has been used by a number of

investigators (3,14,15,17) to quantify the

extent of saturation experienced by agiven tissue in an MT pulse sequence. The

MT ratio indicates the percentage of signal

loss that occurs during the irradiation of

the immobile pool of protons and, conse-quently, is proportional to the pseudo-

first-order rate constant for saturation

transfer between two chemical species of

Fors#{233}n and Hoffman (6). In general, the

MT Ratios for Cranial Structures on MR Images Obtained with Different Sequences

Sequence

Proton- Gadolinium-enhancedStructure T2-weighted density-weighted Ti-weighted Ti-weighted

Cerebrospinal fluidScalp fatWhite matter

0.0 ± 0.90.0 ± 1.2

27.0 ± 0.7

-1.0 ± 1.0 -1.0 ± 2.09.0 ± 0.7 1.0 ± 0.8

28.0 ± 0.7 15.0 ± 0.8

0.0 ± 2.3

0.0 ± 1.016.0 ± 1.1

Caudate 22.0 ± 0.6 23.0 ± 0.9 7.0 ± 1.1 8.0 ± 1.4

Putamen 20.0 ± 1.0 20.0 ± 1.0 8.0 ± 0.8 8.0 ± 1.0Pituitary gland Not measured Not measured 10.0 ± 1.0 2.0 ± 1.4

V

Note-All measurements are MT ratio in percentage ± standard error. Larger MT ratios implygreater suppression and, therefore, decreased signal intensity.

L �

‘a

l�.

b.a.

�:

4�Figure 3. MR images of a 42-year-old female patient. (a) Precontrast Ti-weighted image ob-

tamed without MT shows the white matter with higher signal intensity than the gray matter.

0’) Precontrast Ti-weighted image obtained with MT shows the putamen and caudate headwith higher signal intensity than the white matter. (c) Postcontrast Ti-weighted image ob-tamed without MT shows enhancement of choroid plexus and internal cerebral veins in addi-

tion to the higher signal intensity of white matter compared with gray matter. (d) PostcontrastTi-weighted image obtained with MT shows more conspicuous enhancement of choroidplexus, internal cerebral veins, subependyrnal veins, and cortical vessels, and higher signalintensity of the putamen and caudate head compared with white matter.

larger the MT ratio, the more suppressed

(with lower signal) the tissue will be. Sta-tistical analysis was performed with two-tailed t tests to compare MT ratios of pairsof structures, and a one-way analysis ofvariance with Bonferroni corrections wasused for multiple comparisons of MT ra-tios.

RESULTS

The appearance of the brain on

spin-density- and T2-weighted im-ages when MT pulses were used dif-fered slightly from that expected on

conventional spin-echo images (Fig2). Specifically, MT pulses reduced the

signal from brain by 20% or more buthad little effect on the signal fromscalp fat or cerebrospinal fluid (Table).As a result of this differential suppres-sion of signal, cerebrospinal fluid andfat appeared relatively brighter than

brain on MT spin-echo images whencompared with conventional (ie, non-MT) spin-echo images. Within thebrain itself, we could demonstrate no

visually apparent differences in theusual appearance of intrinsic struc-tures on long repetition time MT im-ages. Specifically, no statistically sig-nificant differences in suppressionwere noted between white matter

and gray matter.On Ti-weighted images, however,

qualitative differences in tissue con-

trast were more dramatic between the

MT and conventional spin-echo im-ages (Figs 3-5). As with the long rep-etition time MT sequences, the MRsignals from all brain structures weredecreased compared with fat, whichtherefore appeared to be abnormally

bright. Moreover, even though brain

signal was globally decreased, somebrain structures were suppressedmore than others. In particular, thewhite matter had the largest MT ra-tios and hence the greatest degree

of signal suppression. As a result,certain brain components (gray mat-ter of the central sulcus, putamen,

caudate, aqueduct, and substantianigra) were rendered significantlybrighter than white matter and be-came conspicuous when MT pulseswere used.

After administration of gadopen-tetate dimeglumine, several more dif-ferences were noted between the con-ventional and MT-saturated Ti-weighted images. Structures thatnormally enhance with gadolinium(eg, choroid plexus, dural sinuses, pi-tuitary, pineal) were rendered evenmore conspicuous than normal whenMT pulses were used. Multiple smallcortical veins, not seen to enhanceappreciably on the conventional Ti-

weighted images, were well visual-ized on the MT images. Quantitativeanalysis of the MT ratios of pituitarytissue revealed a statistically signifi-cant reduction in MT saturation after

administration of contrast material(P < .05).

No significant differences werenoted for the MT ratios of fat or cere-brospinal fluid on Ti- or T2-weightedimages. Both white matter and graymatter structures demonstrated sig-nificantly greater MT effect (greaterMT ratios) on T2-weighted imagesthan on Ti-weighted images(P < .01).

a. b.

Figure 4. MR images of a 31-year-old female patient. Compared with (a) the Ti-weightedimage obtained without MT, (b) the Ti-weighted image obtained with MT demonstrates rela-

tive hyperintensity of gray matter adjacent to the central sulcus (arrows).

DISCUSSION

544 #{149}Radiology February 1994

MT imaging is a relatively newtechnique in which image contrast isthought to be manipulated throughthe selective saturation of a pool ofprotons on proteins, cell membranes,and other macromolecules that are inclose thermal contact with a sphere ofassociated water molecules (i-3).

These macromolecular protons werelong considered to be invisible onclinical MR images because of theirextremely short T2 values (which are

measured in microseconds ratherthan in milliseconds) (23). Since theMR line width is inversely propor-tional to the T2 value, the macromo-lecular pool has a broad spectral den-sity and can thus theoretically bestimulated over a wide range of radiofrequencies. According to currenttheory, by selectively stimulating themacromolecular reservoir, saturationeffects can be transferred to the asso-ciated water pool, thereby affectingthe observed MR signal (7,8,24). Pre-sumably this modulation of the watersignal takes place with dipolar cross-coupling interactions and chemicalexchange processes (2). However,much of the current MT theory re-mains highly speculative; othermechanisms (eg, water interactionswith paramagnetic free radicals onprotein surfaces, spin-lock relaxationeffects, or chemical exchange reac-tions with a hot environment [theMcConnell effect {25}]) may also beinvoked to account for some of theeffect of saturation on contrast ob-served in MT imaging.

Several methods of producing MTsaturation have been proposed, in-cluding use of continuous-wave irra-diation and pulse techniques (5). Be-cause of considerations about specificabsorption rates, continuous-wavemethods are limited to use in low-field-strength systems (2,26,27). Appli-cation of the MT technique at high

field strengths has generally involveduse of pulse methods with which anMT saturation pulse of relatively lowpower is applied with a frequencyoffset of between several hundredand several thousand hertz from the

water resonance. This off-resonancepulse is designed to have sufficientpower to saturate protons in the im-mobile pool but with a frequencyspectrum that will not directly stimu-late protons in the free tissue water.

Use of the MT saturation pulse lim-its the time available for multisectionacquisition and, hence, the number ofallowed sections for a given repetitiontime. In addition, the number of sec-

tions acquired may have to be furtherreduced to avoid exceeding limita-tions of specific absorption rates fortissue energy deposition. The precisepenalty in multisection imaging de-pends on the repetition time: At 600msec, for example, the number ofavailable sections in our study wasreduced by 40%, necessitating twoseparate acquisitions to cover the en-tire brain axially.

Preliminary research by severalgroups of investigators has demon-strated the usefulness of MT imagingin a variety of radiologic applications.Outside the nervous system, MT tech-niques have been used to image theknee (28), heart (29), liver (30), andbreast (3i). Many imaging applica-tions for the central nervous systemhave also been proposed, the mostimportant of which include suppres-sion of background brain tissue at MRangiography (9-li), detection of de-myelinating and degenerative lesions(14-16), and increased visibility ofgadolinium enhancement (18-22).These applications of the MT tech-nique encompass a wide range of im-aging protocols; MT pulses will likelyprove useful in conjunction with Ti-,T2-, and spin-density-weighted imag-ing. To interpret MT-enhanced im-ages of the brain properly, therefore,the radiologist must be aware of theexpected patterns of tissue contrastlikely to be encountered and howthese patterns differ from those seenon conventional spin-echo images.Herein lies our motivation and justifi-cation for performing the present in-vestigation.

Although the exact mechanisms

responsible for the generation of MTcontrast remain unclear, a few simpleconcepts can reasonably explainmany of the signal changes we ob-served in routine cranial imaging.First, MT effects should be most no-ticeable in tissues in which the inter-action between water and macromol-ecules are the principal determinantof relaxation time. This is clearly thecase for skeletal muscle and brain, forwhich several investigators (14,28,29)have measured MT ratios between30% and 80%. Conversely, fluids withrelatively low concentrations of mac-romolecules (eg, cerebrospinal fluidand blood) should experience little ifany suppression of signal when MTpulses are applied. Our measure-ments of MT ratios in these tissues(Table) confirm this hypothesis.

The MT ratios we measured shouldnot be taken as absolute values butrather as indexes of the relative sup-pression of tissues for a given pulsesequence and MT saturation method.Other investigators, with use of dif-ferent MT pulses and imaging proto-cols, have reported different MT ra-tios for cranial tissues (14,20,27,29).Even in our own data, the MT ratiosfor brain tissue were two to threetimes larger for T2-weighted than forTi-weighted images. The precise ex-planation for this effect is unclear butmay be a result of the fact that Ti

weighting usually counteracts thecontrast created with MT pulses. Inaddition, the MT ratio may be in-creased because of the increasednumber of sections acquired (withmore nonselective MT pulses appliedper image) and because of the off-

r

A

‘.

(I

� � � - � � �-‘�

1#{149}).. .J�_1__._. �

�i_ 5%

�s) �

a. b.

ci

Figure 5. MR images of a 56-year-old female patient. Compared with (a) the Ti-weighted image obtained without MT, (b) the Ti-weighted

image obtained with MT shows that the substantia nigra (black arrows) and the periaqueductal gray matter (open arrow) are relatively hyper-

intense.

resonance effects of 180#{176}refocusingpulses used in the multiecho longrepetition time sequence. Since twiceas many refocusing pulses are usedon a double-echo sequence comparedwith a single-echo sequence, onewould expect a greater MT effect onthe double-echo sequence. This MTeffect of multiple 180#{176}pulses has beendocumented in fast spin-echo imag-ing (32). An increase in MT effects inmultisection (compared with singlesection) MR imaging has also beenreported (33). The MT ratios we mea-sured are proportional to and consis-tent with ratios measured in other

studies and allow us to understand ata quantitative level the contrast differ-ences between tissues that we observevisually.

The poor suppression of fat signalwith MT pulses has been noted inseveral previous studies (11,27,29), inwhich computed MT ratios have notusually exceeded 5% . Adipose tissuecontains numerous large intracellulardeposits of lipid stored principally inthe form of medium- and long-chaintriglycerides. These triglycerides areof moderate size and mobility andhence have T2 relaxation times com-parable to those of many other bio-logic tissues. Stored lipids, therefore,may not be considered to be macro-molecules in the sense of being di-rectly saturated with MT techniques.Furthermore, because stored lipidsare hydrophobic and are loculatedinto self-contained droplets withinadipose cells, stored fats have few di-rect interactions with free water.

MT effects are also minimizedwhenever high concentrations of aparamagnetic substance are presentwithin tissue (2,18-22). Because con-trast enhancement at MR imaging isdue to a direct water-gadolinium ioninteraction (34) (not to macromolecu-lar cross relaxation), MT pulses willpreferentially suppress the signalfrom background tissues and, hence,render gadolinium-enhanced areasmore conspicuous. This use of MTpulses has been exploited at low fieldstrength to increase the conspicuity ofa variety of contrast-enhancing le-sions (i8,i9). That such a phenom-enon occurs is confirmed by our mea-surements of the pituitary gland andby our nonquantitative assessment ofother structures such as veins, cho-roid plexus, and other tissues thatnormally enhance.

Within the brain parenchyma itself,we have discovered that a variety ofstructures are more conspicuous onTi-weighted images when MT pulsesare used. The gray matter along thecentral sulcus, the basal ganglia, andsubstantia nigra are the most dramaticexamples of this phenomenon. Oursignal intensity measurements revealthat the gross MR signals from thesestructures are actually reduced by theMT pulses. However, the signals fromthese gray matter structures are lessreduced than are those arising fromthe adjacent white matter; therefore,the gray matter structures are ren-dered more conspicuous on Ti-weighted MT images. The reason isunknown for the differences in MT

ratio between white matter and graymatter and between different graymatter structures. The degree of my-elination, iron deposition, and densityof neurons may all play a role. We arenow conducting research to deter-mine how these effects vary with theage of the patient and with the typeof MT pulse employed.

This pattern of relative prominenceof the central sulcus and of the nucleiof deep gray matter must be recog-nized as a normal finding in the inter-pretation of Ti-weighted MT images.When MT pulses are used in conjunc-tion with gadolinium enhancement,the high signal from these gray matterstructures must not be misconstruedas indicating a pathologic condition.

The prominence of the central sul-cus is a curious and unexplained ob-servation; one immediate practicalbenefit of this phenomenon is that itprovides a method with which to de-termine easily the exact location ofthis sulcus on Ti-weighted MT im-ages. Potentially, this may prove help-ful in at least three situations: (a) when

general radiologists are not familiarenough with normal gyral anatomy tolocate this structure with traditionalmeans; (b) when coronal, oblique, orlow axial images are obtained (wheneven expert neuroradiologists mayfind it difficult to locate the sulcus);and (c) whenever normal gyral land-marks are distorted by tumors,masses, or surgical change.

We conclude that in routine cranialimaging, MT saturation pulses pro-duce new patterns of tissue contrast

546 #{149}Radiology February 1994

that differ substantially from thoseexpected on conventional spin-echoimages. These patterns must be recog-nized for proper interpretation of pre-and postcontrast MR images thathave been obtained with MTpulses. #{149}

References1. Wolff SD, Balaban RS. Magnetization

transfer contrast (MTC) and tissue waterproton relaxation in vivo. Magn ResonMed 1989; 10: 135-144.

2. Balaban RS, Ceclder IL. Magnetization

transfer contrast in magnetic resonanceimaging. Mag Reson Q 1992; 8:116-137.

3. EngJ, Ceckler TL, Balaban RS. Quantita-tive ‘H magnetization transfer imaging invivo. J Magn Reson 1991; 17:304-314.

4. HajnalJV, Baudouin CJ, Oatridge A,

Young IR, Bydder GM. Design andimplementation of magnetization transferpulse sequences for clinical use. J ComputAssist Tomogr i992; i6:7-i8.

5. Schneider E, Prost RW, Glover GH.Pulsed magnetization transfer versus con-

tinuous wave irradiation for tissue contrastenhancement. JMRI 1993; 3:417-423.

6. Fors#{233}n5, Hoffman RA. Study of moder-ately rapid chemical exchange reactions bymeans of nuclear magnetic double reso-nance. J Chem Phys 1963; 39:2892-290i.

7. Koenig SH, Bryant RG, Halenga K, JacobGS. Magnetic cross relaxation among pro-tons in protein solution. Biochemistry 1978;17:4348-4358.

8. Edzes HT, Samulski ET. Cross relaxationand spin diffusion in the proton NMR ofhydrated collagen. Nature 1977; 265:521-523.

9. Pike GB, Hu BS, Glover GH, Enzmann DR.Magnetization transfer time-of-flight mag-netic resonance angiography. Magn ResonMed 1992; 25:372-379.

10. Edelman RR, Ahn 55, Chien D, et al. Im-proved time-of-ffight MR angiography ofthe brain with magnetization transfer con-trast. Radiology 1992; 184:395-399.

ii. Lin W, Tkach JA, Haacke EM, Masaryk TJ.Intracranial MR angiography: application

of magnetization transfer contrast and fatsaturation to short gradient-echo, velocity-

compensated sequences. Radiology 1993;186:753-761.

12. Wolff SD, EngJ, Balaban RS. Magnetiza-tion transfer contrast: method for improv-ing contrast in gradient-recalled-echo im-ages. Radiology 1991; 179:133-137.

13. Ceckler TL, Kanno K, Kador PF, BalabanRS. Magnetic resonance imaging of therabbit eye: improved anatomical detail us-

ing magnetization transfer contrast. InvestOphthalmol Vis Sri 1991; 32:3109-3113.

14. Dousset V, Grossman RI, Ramer KN, et al.Experimental allergic encephalomyelitisand multiple sclerosis: lesion characteriza-tion with magnetization transfer imaging.Radiology 1992; 182:483-491.

is. Lexa FJ, Galetta SL, Grossman RI. Phasicchanges in magnetization-transfer rate inhuman visual pathways after unilateralenucleation (abstr). Radiology 1992; 185(P):i99.

16. Chew WM, Rowley HA, Barkovich JA.Magnetization transfer contrast imaging inpediatric patients (abstr). Radiology 1992;185(P):281.

17. Ordidge RJ, Helpern JA, Knight RA, Qmg

z, Welch KMA. Investigation of cerebralischemia using magnetization transfer con-trast (MTC) MR imaging. Magn Reson Im-aging 1991; 9:895-902.

18. Tanttu JI, Sepponen RE, Lipton MJ, Kuu-sela T. Synergistic enhancement of MRIwith Gd-DTPA and magnetization transfer.

J Comput Assist Tomogr 1992; 16:19-24.19. Kurki TJI, Niemi PT, Lundbom N. Gado-

linium-enhanced magnetization transfercontrast imaging of intracranial tumors.JMRI 1992; 2:401-406.

20. Finelli DA, Hurst GC, Gullapali RP, BellonEM. Improved contrast of enhancingbrain lesions on postgadolinium, Ti-weighted spin-echo images with use ofmagnetization transfer. Radiology 1994;190:553-559.

21. Mathews VM, KingJC, Elster AD, Hamil-ton CA. Cerebral infarction: effects ofdose and magnetization transfer contrast atgadolinium-enhanced MR imaging. Radiol-ogy 1994; 190:547-552.

22. Elster AD, KingJC, Mathews VP, Hamilton

C. Improved detection of gadolinium en-hancement using magnetization transferimaging. Neuroimaging Clin North Am (in

press).23. Bottomley PA, Foster TH, Argersinger RE,

et al. A review of normal tissue hydrogen

relaxation times and relaxation mecha-nisms from 1-100 MHz: dependence ontissue type, NMR frequency, temperature,species, excision, and age. Med Phys 1984;11:425-449.

24. Koenig SH, Brown RD III, Ugolini R.Magnetization transfer in cross-linked bo-vine serum albumin solutions at 200 MHz:a model for tissue. Magn Reson Med 1993;29:311-316.

25. McConnell HM. Relaxation rates bynuclear magnetic resonance. J Chem Phys1958; 28:430-431.

26. Niemi PT, Komu MES, Koskinen 5K. Tis-

sue specificity of low-field-strength magne-tization transfer contrast imaging. JMRI1992; 2:197-201.

27. Swallow CE, Kahn CE Jr, Halbach RE,Tanttu JI, Sepponen RE. Magnetizationtransfer imaging of the human leg at 0.1 T:preliminary study. Magn Reson Imaging

1992; 10:361-364.28. Wolff SD, Chesnick 5, FrankJA, Lim KO,

Balaban RS. Magnetization transfer con-trast: MR imaging of the knee. Radiology

1991; 179:623-628.29. Balaban RS, Chesnick 5, Hedges K, Samaha

F, Heineman FW. Magnetization transfercontrast in MR imaging of the heart. Radi-ology 1991; 180:671-675.

30. Outwater E, Schnall MD, Braitman LE,

Dinsmore BJ, Kressel HY. Magnetizationtransfer of hepatic lesions: evaluation of anovel contrast technique in the abdomen.Radiology 1992; 182:535-540.

31. Pierce WB, Harms SE, Flamig DP, GriffeyRH, Evans WI’, Hagans JE. Three-dimen-sional gadolinium-enhanced MR imagingof the breast: pulse sequence with fat sup-pression and magnetization transfer con-trast-work in progress. Radiology 1991;

181:757-763.32. Listerud J, Einstein 5, Outwater E, Kressel

HY. First principles of fast spin echo.Magn Reson Q 1992; 8:199-214.

33. Dixon WT, Engels H, Castillo M, Sardashti

M. Incidental magnetization transfer con-trast in standard multislice imaging. Magn

Reson Imaging 1990; 8:417-422.34. Wood ML, Hardy PA. Proton relaxation

enhancement. JMRI 1993; 3:149-156.