Download - Week 1. Basics of multimodal imaging and image processing. Functional magnetic resonance imaging
2012.10.30.
1
Multimodal Imaging in Neurosciences Course
Introduction to Multi-modalneuroimagingDr. Ervin Berenyi, MD, PhDDr. András Jakab, MD, PhDDr. Peter Katona, MD
Diagnostic neuroimaging modalities
CT – Computed TomographyBrain anatomyStereotactic reference frame
Structural MRIFine brain anatomyVascular structure
Diffusion, perfusion MRIFine pathologicalinformation
Intra-operative imagingmodalities, open MRI, low-field
Multi-modal imagingspectrum for
MR SpectroscopyBrain metabolismBiochemical mapping
Positron EmissionTomography PETBrain metabolismBrain function
Functional MR imaging fMRIBrain function
Electro encephalography, LORETTA, Magnetoencephalography
1.Diagnostic imaging2.Research
3.Neurosurgery
What is multimodality?Combining images and information from multipleimaging tools, devicesAnatomical alignment of imagesFusion display, co-analysis of multipleinformation sources
What is needed for multimodality?What is needed for multimodality?CT, PET, MRI, SPECT, EEG, …Hybrid devices – PET-CT, PET-MRIImage processing skills to create image fusions, etc.
PET-CT HYBRID
CT: anatomy + attenuationcorrectionPET: metabolism, function
PET-MRI HYBRID SCANNER
Acquire PET and MRI togetherGreat technological challenge$$$
MeasuringMeasuring tissuetissue propertiesproperties withwith MRIMRI
T1 relaxation
T2 relaxation
Proton density
Ti diff i
Structural MRI
Diffusion-weightedTissue diffusion
Diffusion direction
Diffusion anisotropy
Diffusion maps
Metabolites
weightedimaging
Diffusion tensorimaging
Diffusion spectralimaging, HARDI
MR spectroscopy
2012.10.30.
2
VISUALIZATION OF STRUCTURE
Recidive tumor, 2 foci, purple and magenta
Markers on the skin
removed temporal lobe parts
VISUALIZATION OF FIBERS
OPTIC RADIATION
CORTICOSPINAL TRACT
Functional MR ImagingDr. Ervin Berenyi, MD, PhDDr. András Jakab, MD, PhDDr. Peter Katona, MD
Multimodal Imaging in Neurosciences Course
Part I.Basics of fMRI and functionalmappingpp g
Brain functions – how to interpret
The synchronous activity of neuronal groupsCerebral cortexExamples of brain functions
Visual processingAuditory processingMemory functions, recallWernicke areaWernicke areaBroca areaMovement of limbsEmotional response: e.g. human face
„not processing anything” -default mode networks andresting state networks
COGNITIVE PROCESSING IN THE BRAIN
Primary sensory areas (somato-, auditory, etc.)Secondary, tertiary, etc. sensory areas (i.e. visual: 5-9
levels)Association areas„Association areas for higher cognitive functions”Motor response behavior
+ Parallel processing (notpurely hierarchical!)
Motor response, behavior
Somatosensory cortex (SI)
Somatosensory cortex (SII)
Parietal association area
„DLPFC – higher cognitive processing”
Drive, behavioral processing etc.
Speech motor center
2012.10.30.
3
The brain never rests!Default network
Default mode networkDefault state networkTask-negative network
„Wandering and Wondering”Posterior cingulate cortexPrecuneusPrefrontal cortex
DaydreamingSynchronised areasAge dependencyDiseases affecting itNot dreaming!
Fair DA, Cohen AL, Power JD et al. (2009). "Functional brain networks develop from a 'local to distributed' organization". PLoS Comput Biol 5 (5): e1000381
Mapping neuronal function
Electric activity of neuronsAction potential, propagation of signalElectric current – magnetic field variations
Metabolic activity of neurons
Electro encephalography EEGMagnetoencephalography MEG
Positron emission tomographyGlucose metabolism
Blood supply of neuronsVasodilatation, perfusion change
Rapid changes of cell compartmentsCell swelling?
(18F-FDG) PET
fMRI
fDTI
History“[In Mosso’s experiments] the subject to be observed lay on a delicately balanced table which could tip downward either at the head or at the foot if the weight of either end were increased. The moment emotional or intellectual activity began in the subject, down went the balance at the head-end, in consequence of the redistribution of blood in his system.”
-- William James, Principles of Psychology (1890)
Angelo Mosso(1846 1910)(1846-1910)
E = mc2
???Zago et al. (2009) The Mosso method for recording brain pulsation: The forerunner offunctional neuroimaging. Neuroimage
HistoryThe first evidence for the coupling between energymetabolism and brain blood perfusion (animals)The blood volumen elevated during brain activitySir C. S. Sherrington, 1890Seymour Kety & Carl Schmidt, 1948
Increased oxigen take-upDilatation of blood vessels
Near infrared spectroscopy
Sir Charles Scott Sherrington
(1857-1952)ea a e spec oscopyPETfMRI (90’s): Seji Ogawa, Ken Wong
(1857 1952)
Cerebral Cortex. 12:225-233; 2002.
2012.10.30.
4
Activity Increases Flow
• sensory stimulation leads to increased blood flow
• sciatic nerve, electronic stimulation (0,2 V 5-10 Hz), rats, automated video dimension analyzer
Blood pressure
dimension analyzerArteriole diameter Blood velocity
Data Source: Ngai et al., 1988, Am J PhysiolFigure Source, Huettel, Song & McCarthy, Functional Magnetic Resonance I i
Summary of in vivo imaging methods
Structural imagingCTMRI
T1 – 3DT1 – „anatomical”T2FLAIR, DWI, etc.
Functional imagingPETfMRI
…..
fMRI
Structural MRI Functional MRI
Good spatial resolution = 0.6 – 1 mm
Short scan time (a few minutes)
One time point is imaged
Good tissue contrast
No image post-processing is required
The result is robust
Bad spatial resolution = 2 – 4 mm
Long scan time (10-30 minutes)
Multiple time points, multiple scans
Bad tissue contrast
Post-processing is required
The result depends on the patient, theprotocol and paradigm
OK. Now show me the trick.
2012.10.30.
5
25
-Four globin chains-Each chain contains a haem molecule
-Each haem has an iron atom in the center(Fe)
-Each haem can absorb one oxygen
The hemoglobine
Source: http://wsrv.clas.virginia.edu/~rjh9u/hemoglob.html, Jorge Jovici& Huettel, Song, McCarthy, Functional Magnetic Resonance Imaging
molecule (O2)
-oxy-Hgb (four O2) has DIAMAGNETIC effect →it does not affect the magneticfield ΔB
-deoxy-Hgb is PARAMAGNETIC → if[deoxy-Hgb] ↓ → then local ΔB ↓
Diamagnetism and paramagnetism
Diamagnetism(oxy- & carbonmonoxyhemoglobine)No magnetic momentumHas paired electrons
Paramagnetism (deoxyhemoglobine)Magnetic momentum – atoms behave as small magnetsHas unpaired electrons
Measuring deoxy-hemoglobine• During fMRI acquisitions, we get information of the brain’s deoxy-
hemoglobine content
• The relative oxygenation changes with the deoxygenated hemoglobinecontent
Seiji Ogawa
How does this work? The BOLD effect!Blood Oxygen Level DependentThe funcitonal activity is coded in the BOLD effect.
OxyHb and DeoxyHb- their MR relaxation properties are different!
deoxyHb: paramagnetic!!!
M
Source:, Huettel, Song & McCarthy, 2004, Functional Magnetic Resonance Imaging
time
MxySignal
Mo sinθT2* task
T2* control
TEoptimum
Stask
ScontrolΔS
Source: Jorge Jovicich
HEMODYNAMIC RESPONSE
2012.10.30.
6
End of Part I. – any questions?
Part II.How to perform an fMRI?p
The MRI recipe
Human, made of
1. Patient (water + fat = lot of spins)
2. Excite (Shout at the patient with a radiofrequency coil)
3. Wait until the excited spins „relax”
4. During relaxation, the spins (water + fat =patient) shout back at you, theysend an ECHO
5. You listen to the echo and record it(this is the k-space acquisition)
Repeat this! This is called SEQUENCE
6 D d th i l t i !,excitable spins (H
proton spins)ECHO
6. Decode the signal, get image!
MRI sequencesImage coded as waves, Fourier transformation is used to „decode” the rawsignal and get an image
You can „excite” the spin system in numerous ways to have image signals, i.e. SPIN ECHO or GRADIENT ECHO sequences.
GRADIENT ECHO SEQUENCES ARE SENSITIVE FOR DEOXYHEMOGLOBINE CHANGES!
How does echo planar imaging works?Echo-planar imaging (SE-EPI, GRE-EPI)T2 contrastAfter one excitation, an entire slice is read out.It is a fast MR imaging sequenceHas many artifacts, i.e. susceptibility
IMAOIS – www.imaios.com
2012.10.30.
7
How to perform an fMRI scan? Checklist!Can our MRI device perform fast EPI, what is thefield strength? 1.5T vs. 3T?What are we interested in?
fMRI experiments are task-specificIt is necessary to construct a PARADIGM which „observes” one specific brain function
D h i i kill ?Do we have image processing skills?$$$Patient cooperative?IQ, attention?Do we have enough time?Sedation, drugs, etc.
fMRI and all the tools
The first step: imaging the anatomyT1 weighted anatomical images as references
• High resolution images (1x1x2.5 mm)• 3D acquisition• pl. 64 anatomical images ~ 5 perc
Anatomical acquisition
Slice Thicknesse.g., 6 mm
SAGITTAL SLICE IN PLANE SLICE
VOXEL(Volumetric Pixel)
3 mm 6
mm
In-plane resolutione.g., 192 mm / 64= 3 mm
Number of Slicese.g., 10
IN-PLANE SLICE
Field of View (FOV)e.g., 19.2 cm
3 mm
Matrix Sizee.g., 64 x 64
Second step: the actual fMRI acquisitionT2*-weighted images
• Image contrast relates to neuronal activity• Low spatial resolution (3x3x5 mm)• One volume of the brain is acquired in 2 seconds!• We acquire many volumes in time (4D), ie. 150• Repeated scanning
first volume(2 sec to acquire)
…
Paradigm and block design
~2 sec
Functional images
Time
fMRIsignal(% change
ROI Time Course
Tasks
Statisticalactivation map on T1 image
Time
~ 5 minutesRegion of interest kijelölés (ROI)
2012.10.30.
8
Interpreting fMRI results: LOCALIZATION
TALAIRACH ATLAS- 1988- 1 SZEMÉLY
Variability of sulci - problematic
Source: Szikla et al., 1977 in Tamraz & Comair, 2000
Fathers of Localization (brain atlases)
Jean Talairach(January 15, 1911, Perpignan
– March 15, 2007, Paris)
Gabor Szikla
Anatomical localization of activity: gyri and sulci
gray matter (dendrites & synapses)
white matter (axons)
ANK
FUNDUS
BA
FISSURE
Source: Ludwig & Klingler, 1956 in Tamraz & Comair, 2000
How to display fMRI results?
Creating 3D visualizations of the individual brain: Skull-stripping, inflating the cortex
Brain extraction Inflation
2012.10.30.
9
Segmentation, filtering, masking
Fuzzy thresholding Anisotropic filtering Only brain
Standardization of fMRI images to brainatlases
Displaying fMRI fMRI display
End of Part II. – any questions?
Part III.Examples and research applicationsp pp
2012.10.30.
10
What functions can we image usingfMRI?
Paradigm-dependent!Vision („vibrating checkboard”)Audition (variable frequency stimuli)Limb movement – activePassive limb movement - infantsMemory (hometown walking test)Speech… and many others (but not everything!)
The logic of a „simple” fMRI experimentRest = empty screen
Task1 Task 2Time
The subject views an object, i.e. apple „Scrambled” – image
Results: object recognition
Kalanit Grill-Spector et al.
First images of visual activityFlickering CheckerboardOFF (60 s) - ON (60 s) -OFF (60 s) - ON (60 s) - OFF (60 s)
Source: Kwong et al., 1992
CO-ACTIVATION OF V1 -> V2.. AFTERVISUAL STIMULUS
Motor paradigm of the left hand
2012.10.30.
11
Finger tapping test of theright hand
Source: Katona P., DEOEC
Lesion in the left precentral gyrus (malformation) – REDHand movement activation: Yellow, CS tract: yellow
Jakab, Katona et al.
HOMUNCULUS Left hand
Source: Berenyi, Emri, Jakab et al
Left foot
Forrás: Berényi E, Emri M. DEOEC
Auditory activation
Task:
Listening
to
ordersorders
Forrás: Berényi E, Emri M. DEOEC
2012.10.30.
12
FREQUENCY PROGRESSION OF HUMAN AUDITORY CORTEX
J Neurophysiol. 91:1282-1296, 2004.
Late speech development – pathologicallocalization of speech centers?
Radiology. 2003;229:651-658.
Speech paradigm: say a word beginningwith a,b,c, etc.
Jakab A, Katona P et al.
Localizing swallowing movement
AJNR. 20:1520-0526. 1999.
2012.10.30.
13
2012.10.30.
14
Szentágothai TK Szentágothai TK --Semmelweis EgyetemSemmelweis EgyetemMR KutatóközpontMR Kutatóközpont
fMRI in afMRI in a Case of Childhood EpilepsyCase of Childhood Epilepsy
Lajos R KozakMR Research Center, Semmelweis University, Budapest, Hungary
Patient historyPatient historyA case of drug resistant epilepsy A case of drug resistant epilepsy
8 yrs old right handed boyBorn on term from uneventful pregnancy
First seizures at 3.5 yrsAbout the time of falling asleep starting with left hand twithcing then generalizingLater atypical absence seizures
EEG resultsEEG resultsNormal EEG on the onsetLater slow spike and wave activity developed with clinical abscenceFinally, electric status epilepticus during sleep (ESES), irregular high amplitude spike and wave activity, during the whole night
Physical examinationParesis on the left limbs
Patient historyPatient historyImaging
Smaller right hemisphere
On T1 weighted images (A-B) widespread irregularities of the cortical surface suggestive of multiple small folds with abnormally thick cortex, irregular appearance of the gray
tt hit tt j timatter-white matter junction
suggestive of polymicrogyria
On FLAIR images (C)numerous high intensity foci predominantly in the subcortical white matter
Question: is the malformed cortex functional?
Kozák et al., Clin Neurosci2009;62(3–4):130–135.
fMRI #fMRI #11no result
Imaging at 3T Philips Achieva scannerTR=3000ms, TE=30ms, FA=75°, 3x3x3mm2 voxels (80x80 matrix, 240x240 FOV), axial slices, no gap, SENSE factor of 2Block design paradigm, WHAT WAS THE WHAT WAS THE WHAT WAS THE WHAT WAS THE Block design paradigm, 24s movement, 24s rest
• flexion/extension of fingers ~0.5-1Hz
• left and right limb moved in separate blocks
PROBLEM WITH THE PROBLEM WITH THE fMRI?fMRI?
PROBLEM WITH THE PROBLEM WITH THE fMRI?fMRI?
movementrest
Bad acquisition ?Bad stimulation ?Overanesthetized ?
fMRI #1fMRI #1The reason for unsuccesful fMRI?
500-700μV
Electric status epilepticus during sleep (ESES) ?
Clonazepam was the solution
Kozák et al., Clin Neurosci2009;62(3–4):130–135.
2012.10.30.
15
Preop.
fMRI #fMRI #22right hand movement pre- and postoperatively
Postop
Preop.
fMRI #fMRI #22left hand movement pre- and postoperatively
Postop
Functional reorganization to the healthy hemisphereFunctional reorganization to the healthy hemisphere
ConclusionsConclusions
Passive rangePassive range--ofof--movement paradigms are movement paradigms are considered useful considered useful for the mapping of sensory-motor cortex in pediatric epilepsy patientsin pediatric epilepsy patients.
If fMRI failsfMRI fails in this patient population we have toIf fMRI fails fMRI fails in this patient population we have to check if there is ongoing epileptic activity if there is ongoing epileptic activity during anesthesia
These paradigms are able to describe cortical able to describe cortical reorganizationreorganization, thus they have clear have clear prognostic value in a preprognostic value in a pre--operative settingoperative setting.
Research with fMRI
Summary of facts so far
fMRI is based on the BOLD = BloodOxygen Level Dependent contrastNeurovascular couplingA stringent paradigm is required( l)(protocol)Mapping brain activity can be achieved in living humansMany factors can influence theresultsfMRI = localization
"...the single most critical piece of equipment is still the researcher's own brain. All the equipment in the world will not help us if weequipment in the world will not help us if we do not know how to use it properly, which requires more than just knowing how to operate it. Aristotle would not necessarily have been more profound had he owned a laptop and known how to program. What is badly needed now, with all these scanners whirring away, is an understanding of exactly what we are observing, and seeing, and measuring, and wondering about."
-- Endel Tulving, interview in Cognitive Neuroscience (2002, Gazzaniga , Ivry & Mangun, Eds., NY: Norton, p. 323)
2012.10.30.
16
A new localizationism?The accepted application
Surgical planningFor cognitive neuroscience, localizationitself has INFERIOR significancePopularity, factoid literature
Example for a BAD fMRI experiment
~2 sec
Time
Task 1: Subject observes a car on a screenTask 2: Subject observesnoise (control)Task 3: Subject observesCAR + Elmo Muppet
BAD INTERPRETATION OF FMRI RESULTS CANSTILL MAKE A JOURNAL PUBLICATION?
- =Elmo + CARCAR against noise
Visual areas for „carobservation”
Visual areas for „car + elmoobservation”
Elmo (negativeelmo)
Elmo Brain Area ???
The brain before the fMRI era
Polyak, in Savoy, 2001, Acta Psychologica
grasping
motion near head
motorcontrol
reaching and pointing
touch
retinotopic visual maps eyemovements
executive control
THE BRAIN AFTER FMRI (INCOMPLETE)
moving bodiessocial cognition
faces objectsstatic bodies
motion perception
orientation selectivitymemory
scenes
Basic types of fMRI researchTesting models, theoriesLocalize the activations after stimuliActivating networks after stimuliSpatial encoding of the brain:
Retinotopy, somatotopy, frequencyRetinotopy, somatotopy, frequencycoding
Behavior and cognitionDiseases, i.e. psychiatryInter-species comparisons
2012.10.30.
17
The future of functional brain imaging
3T, 4T, 7T, … ?Ultra-low-field imagingArterial spin labelingFunctional diffusion tensorimaging (Le Bihan)
ULTRA-LOW-FIELD IMAGINGEarth magnetic fieldSQUID MAGNETOMETRY
Los Alamos, USA
The small electric currents of neuronal activity induce changesin the magnetic field, whichinterferes with the Earth’s and imaging can be performed
Arterial Spin Labeling - ASLinversionslab
i i
excitation
inversionxy
z (=B0)
bloodblood
99
• Perfusion: delivery of metabolites (via local blood flow) (BOLD - hemoglobin)
• Arterial Spin Labeling (ASL): invert of in-flowing blood
• IMAGEperfusion = IMAGEuninverted - IMAGEinverted
imagingplane
Arterial Spin Labeling - ASL
100
• Represents an interesting physiological parameter• Quantitative: fit kinetic curve for perfusion in
ml/100g/min• Lower SNR than BOLD• Limited coverage (~5 slices)
Arterial Spin Labeling - ASL
Magn Reson Med, 48:242-254 (2002)
Arterial Spin Labeling - ASL
Magn Reson Med, 48:242-254 (2002)
2012.10.30.
18
Stroke. 2000;31:680-687.
End of Part III. – any questions?
Part IV.The functional brain connectome
Resting state fMRI
Don’t do anything.
Spontaneous synchronity in the brain = low frequency oscillations
<0.1 Hz neuronal activity is present during „rest”Background for continuous sensory processing?What regions are „synced” ?
2012.10.30.
19
Correlated time courses = networks1. Regional slow neuronal activity
Hypothesis: if two neuronal time courses arecorrelated, the regions are interconnected.
2. Regional slow neuronal activity
3. Their correlation (temporal)
THE SHORT HISTORY OFCONNECTOMICS
Theodor MeynertJules DejerineTracing studies„In vivo methods”:Diffusion tensor imagingFunctional MR imagingFunctional Connectivity
This is called FUNCTIONAL CONNECTIVITY
2012.10.30.
20
ModelingModeling thethe brain’sbrain’s connectionsconnections
Brain regions: network nodesStructural OR functional brain connection strength: network edges
Graph-theoretical analysis, a purely mathematical approach
Node (region)
How can information be exchanged among brain regions?
Short path-length, Low degree
Edge(connection)
ModelingModeling thethe brain’sbrain’s connectionsconnections
Brain regions: network nodesStructural OR functional brain connection strength: network edges
Graph-theoretical analysis, a purely mathematical approach
How can information be exchanged among brain regions?
Long path-length, Low degree
ModelingModeling thethe brain’sbrain’s connectionsconnections
Brain regions: network nodesStructural OR functional brain connection strength: network edges
Graph-theoretical analysis, a purely mathematical approach
How can information be exchanged among brain regions?
Short path-length, High degree(low efficiency)
ModelingModeling thethe brain’sbrain’s connectionsconnections
Brain regions: network nodesStructural OR functional brain connection strength: network edges
Graph-theoretical analysis, a purely mathematical approach
Hub
How can information be exchanged among brain regions?
Example of a highlyefficient network
Hub
2012.10.30.
21
ModelingModeling thethe brain’sbrain’s connectionsconnections
Brain regions: network nodesStructural OR functional brain connection strength: network edges
Graph-theoretical analysis, a purely mathematical approach
How can information be exchanged among brain regions?
Example of an inefficientnetwork (almost random)
The internet
WhatWhat is is thethe corticocortico--corticalcortical brainbrain networknetwork likelike??CORTEX Facebook
Small World Networks
Source: Paul Weinstein’s blog
Modha & Singh. Network architecture of the long-distance pathways in the macaque brain. PNAS, 2010
Network Network propertiesproperties of of thethe brainbrain: : normalnormal developmentdevelopment
Network cost Network efficiency
Gong et al. Age- and Gender-Related Differences in the Cortical Anatomical Network. The Journal of Neuroscience 2009; 29: 15684-15693.
cost=sum(wij)
Network Network propertiesproperties of of thethe brainbrain: : normalnormal and and pathologicalpathological
Gong et al. Age- and Gender-Related Differences in the Cortical Anatomical Network. The Journal of Neuroscience 2009; 29: 15684-15693.
Network Network propertiesproperties of of thethe brainbrain: : gendergender differencesdifferences
Network cost Network efficiency
Gong et al. Age- and Gender-Related Differences in the Cortical Anatomical Network. The Journal of Neuroscience 2009; 29: 15684-15693.
cost=sum(wij)
Network Network propertiesproperties of of thethe brainbrain: : correlationcorrelation withwith intelligenceintelligence
Van Heuvel et al. Efficiency of Functional Brain Networks and Intellectual PerformanceThe Journal of Neuroscience, 2009, 29(23): 7619-7624.
Path length negatively correlateswith IQ, especially in the leftfrontal medial cortex
2012.10.30.
22
DetectingDetecting areasareas withwith similarsimilar connectivityconnectivity profilesprofiles
+Exekutívskill
-Exekutívskill
Jakab A et al. Mapping changes of in vivo connectivity patterns in the human mediodorsal thalamus: correlations with higher cognitive and executive functions. Brain Imaging and Behavior 2012; DOI: 10.1007/s11682-012-9172-5
Network Network propertiesproperties of of thethe brainbrain: : schizophreniaschizophrenia
LOSS OF HIERARCHICAL ORGANIZATION IN FRONTAL
REGIONS
Bassett, D. S. et al. 2008
Van Heuvel. et al. 2010
n=9 (HLFA) vs. n=40 (controls)
Network Network propertiesproperties of of thethe brainbrain: : highhigh functioningfunctioning autisticautistic adultsadults
Jakab A, Spisak T, Szeman-Nagy A, Beres M, Molnar P, Emri M, Berenyi E. Pathological patterns of functional connectivity and white matter anisotropy in high functioning autistic adults. Under review @ PLoS One
Suggests the impairmentof long-rangeassociation fibers, especially in theleft fronto-temporo-ocipital connectivities
Thank you for your attention!
Presentation credits:
Dr. András Jakab, M.D. Ph.D.Dr. Ervin Berényi, M.D. Ph.D.Dr. Péter Katona, M.D.Dr. Miklós Emri, Ph.D.Tamás Spisák, M.sc.