Slide 1 Functional Imaging: A review of fmri, DTI and Non-invasive Perfusion Imaging. Slide 2. Slide 3. Overview

Size: px

Start display at page:

Download "Slide 1 Functional Imaging: A review of fmri, DTI and Non-invasive Perfusion Imaging. Slide 2. Slide 3. Overview"

Wilfred Briggs
6 years ago
Views:

1 Slide 1 Functional Imaging: A review of fmri, DTI and Non-invasive Perfusion Imaging Kristine Mosier DMD, Ph.D. Neuroradiology & Imaging Science Department of Radiology Clinical fmri, Chief Head & Neck Imaging Associate Professor of Radiology, Neuroscience and Biomedical Engineering Indiana University School of Medicine & IUPUI Slide 2 Overview Functional Brain Mapping Neurophysiology and hemodynamic basis of BOLD / CBF. fmri paradigms, data acquistion and processing. Clinical case examples. Diffusion Tensor Imaging (DTI) & Fibertracking. Non-invasive Perfusion Imaging (Arterial Spin Labeling). Cases Slide 3

Slide 4 Functional Imaging: fmri Brain activity can be

2 Slide 4 Functional Imaging: fmri Brain activity can be mapped using either BOLD technique (Blood Oxygen Level Dependent) or rcbf. Both BOLD and CBF changes dependent on neurovascular coupling. BOLD signal most closely correlated with LFP (local field potentials). fmri performed in the neurosurgical setting to map eloquent cortex. Slide 5 Slide 6 fmri

Increase in focal oxygenated blood decrease in

3 Slide 7 BOLD mechanism: summary Neuronal activity focal net increase in blood flow and oxygenation. Increase in focal oxygenated blood decrease in deoxyhemoglobin less T2* effect increase in signal intensity. Slide 8 Slide 9

4 Slide 10 BOLD fmri: contrast mechanism Relative mismatch between O 2 delivery and O 2 extraction during activation period Blood flow is increased to activated regions of brain O 2 extraction also increased, but less than increase in O 2 delivery Slide 11 BOLD fmri: contrast mechanism Thus increased oxyhb at post-capillary level decreased deoxyhb DeoxyHb is paramagnetic decreases T2* (decreases signal) Decrease in local deoxyhb results in increased signal intensity on T2*-wtd images ( 1-5%) Slide 12 Basis of BOLD fmri From: Moseley ME & Glover GH. NeuroImaging Clinics of North America; Functional Neuroimaging 5(2): , 1995

161-191, 1995 Slide 14 Raw Image Time Series visual stim no stim visual

5 Slide 13 Basis of BOLD fmri From: Moseley ME & Glover GH. NeuroImaging Clinics of North America; Functional Neuroimaging 5(2): , 1995 Slide 14 Raw Image Time Series visual stim no stim visual stim no stim Slide 15 Difference Image Time Series visual stim no stim visual stim no stim

Slide 16 Peak BOLD signal arises at the level of the

at high field (e.g. 7-9T) within 200 µm of LFP.

6 Slide 16 Peak BOLD signal arises at the level of the postcapillary venule. Problem: contribution to signal from draining veins spatial, temporal artifact. Animal expt. at high field (e.g. 7-9T) within 200 µm of LFP. Humans: several studies BOLD accurate to w/in 1cm of electrode. Slide 17 Slide 18 Echo Planar Imaging FFT k-space image space

Accurate processing: need to remove drift, motion, physiological noise.

7 Slide 19 fmri Data acquistion Acquire time-series of fast images while subject performs sensorimotor, language or cognitive task Process time-series data using statistical methods & compare signal change during task performance with signal during rest / baseline. Accurate processing: need to remove drift, motion, physiological noise. Slide 20 Patient Preparation Paradigm Design Data Acquisition fmri : Study Overview Data Transfer Image Reconstruction and Processing Workstation Statistical Maps Computation Visualization of Maps and Analysis Slide 21

Slide 22 Slide 23 fmri Data: Statistical Analysis Most

8 Slide 22 Slide 23 fmri Data: Statistical Analysis Most commercially available and custom algorithms use GLM (General Linear Model). Y = M*a+e; Y = data vector, M = model of amplitude response, a = response amplitude, e = noise. Current scanner software (GE, Siemens) has real-time processing using t-test: BEWARE; not all noise & artifact removed! Slide 24 fmri Post-Procesing: FT paradigm Raw EPI

$#!/bin/bash # set HOST = `hostname` echo "subj=080821_hpa" read subj echo $subj stadir=/data6/${subj} refpath=/home/yang/fmriref #TR=3s #nslices=32 #frame=144 ##tasks=(ft RB LB VS RM WG NA)$

9 #!/bin/bash # set HOST = `hostname` echo "subj=080821_hpa" read subj echo $subj stadir=/data6/${subj} refpath=/home/yang/fmriref #TR=3s #nslices=32 #frame=144 ##tasks=(ft RB LB VS RM WG NA) tasks=(bwchbd CLchbd) #frames=( ) ##datadir=(8-fmri-epi-ft 22-fMRI-EPI-RB 15-fMRI-EPI-LB 29-fMRI-EPI-WG 36-fMRI-EPI-RM 43-fMRI-EPI-NA) datadir=(10-ep2d_pace_stamp_ba 14-ep2d_pace_stamp_WG 16-ep2d_pace_stamp_NA 18-ep2d_pace_stamp_RM 20-ep2d_pace_stamp_PL 26-ep2d_pace_stamp_CS) cd ${stadir}/afni ## segmentation mprage in SPM5 and generate brain only mask on segemented brain ## coregister T2 or T1 to mprage in SPM5 and then coregister EPI-MC.nii to rt2.nii or rt1.nii ## coregister *MC.nii in SPM5 --- estimate ONLY ##first convert these spm_output_files back to afni #files=`ls r*mc*.nii` #for file in ${files} #do #echo run for ${file} # origfile=${file#r} # \rm tmp* # 3dresample -dxyz prefix tmp-rs -inset ${file} # 3dresample -master tmp-rs+orig -inset ${origfile} -prefix tmp-orig-rs # 3dWarpDrive -affine_general -base tmp-rs+orig -prefix ${origfile%.nii}-fix tmp-orig-rs+orig #done #files=`ls rmask*.nii` #for file in ${files} #do #echo run for ${file} # origfile=${file#r} # \rm tmp* # 3dresample -dxyz prefix tmp-rs -inset ${file} # 3dresample -master tmp-rs+orig -inset ${origfile} -prefix tmp-orig-rs # 3dWarpDrive -affine_general -base tmp-rs+orig -prefix ${origfile%.nii}-fix tmp-orig-rs+orig # ###3dresample -master ${origfile} -inset ${file} -prefix tmp-rs # ###3dWarpDrive -affine_general -base tmp-rs+orig -prefix ${origfile%.nii}-fix -twopass ${origfile} #done ## now analyze each task #echo all tasks are ${tasks[*]} n=0 for task in ${tasks[*]} do echo run for $task 3dvolreg -verbose -Fourier -prefix ${subj}-${task}-mc.nii -base 0 -zpad 4 -tshift 0-1Dfile ${subj}-${task}mc ${task}.nii 1dplot -ps -volreg -one -nopush -xlabel ${subj}-${task} ${subj}-${task}mc > ${subj}-${task}mc.ps 3dNotes -HH " " ${subj}-${task}-mc.nii.gz 3dAutomask -prefix Mask-${task} ${subj}-${task}-mc.nii.gz 3dDespike -prefix ${subj}-${task}-mcds ${subj}-${task}-mc.nii.gz 3dmerge -1blur_fwhm 5 -doall -prefix ${subj}-${task}-mcdssm ${subj}-${task}-mcds+orig file=${subj}-${task}-mcdssm+orig cp ${subj}-${task}mc mc 3dDeconvolve -input ${file} \ done -progress \ -mask Mask-${task}+orig \ -polort A -num_stimts 8 \ -stim_file 1 ${refpath}/test_on.wav.1d -stim_label 1 'On' \ -stim_file 2 ${refpath}/test_off.wav.1d -stim_label 2 'Off' \ -stim_file 3 'mc[0]' -stim_base 2 -stim_label 3 Roll \ -stim_file 4 'mc[1]' -stim_base 3 -stim_label 4 Pitch \ -stim_file 5 'mc[2]' -stim_base 4 -stim_label 5 Yaw \ -stim_file 6 'mc[3]' -stim_base 5 -stim_label 6 ds \ -stim_file 7 'mc[4]' -stim_base 6 -stim_label 7 dl \ -stim_file 8 'mc[5]' -stim_base 7 -stim_label 8 dp \ -num_glt 4 \ -glt_label 1 'On' -gltsym 'SYM: +On' \ -glt_label 2 'Off' -gltsym 'SYM: +Off' \ -glt_label 3 'On-Off' -gltsym 'SYM: +On -Off' \ -glt_label 4 'Off-On' -gltsym 'SYM: +Off -On' \ -bucket ${subj}-${task}+mlr -nocout -tout -fout -rout -vout \ -fitts ${subj}-${task}+fit -errts ${subj}-${task}+ert -xsave \ -cbucket ${subj}-${task}+cbk -x1d ${task} 3drefit -addfdr ${subj}-${task}+mlr+orig #3dFDR -input ${subj}-${task}+mlr+orig -mask_file Mask-${task}-fix+orig -prefix ${subj}-${task}+mlrfdr #3dNotes -HH " " ${subj}-${task}+mlrfdr+orig ((n=n+1)) Slide 25 fmri Post-Procesing: FT paradigm Bas_MoCo Slide 26 Post-Processing: AFNI Slide 27 fmri Post-Procesing: FT paradigm Motion Parameters

Slide 28 fmri Post-Procesing: FT paradigm <matrix # ni_type = "11*double" # ni_dimen = "144" # ColumnLabels = "Run#1Pol#0 ; Run#1Pol#1 ; Run#1Pol#2 ; On#0 ; Off#0 ; Roll#0 ; Pitch#0 ; Yaw#0 ; ds#0 ;

.143" # NRowFull = "144" # RunStart = "0" # Nstim = "2" # StimBots = "3,10" # StimTops = "3,10" # StimLabels = "On ; dp" # Nglt = "4" # GltLabels = "On ; Off ; On-Off ; Off-On" # GltMatrix_000000 =

10 Slide 28 fmri Post-Procesing: FT paradigm <matrix # ni_type = "11*double" # ni_dimen = "144" # ColumnLabels = "Run#1Pol#0 ; Run#1Pol#1 ; Run#1Pol#2 ; On#0 ; Off#0 ; Roll#0 ; Pitch#0 ; Yaw#0 ; ds#0 ; dl#0 ; dp#0" # ColumnGroups = "3@-1,1,6@0,8" # RowTR = "2" # GoodList = "0..143" # NRowFull = "144" # RunStart = "0" # Nstim = "2" # StimBots = "3,10" # StimTops = "3,10" # StimLabels = "On ; dp" # Nglt = "4" # GltLabels = "On ; Off ; On-Off ; Off-On" # GltMatrix_ = "1,11,3@0,1,7@0" # GltMatrix_ = "1,11,4@0,1,6@0" # GltMatrix_ = "1,11,3@0,1,-1,6@0" # GltMatrix_ = "1,11,3@0,-1,1,6@0" # CommandLine = "3dDeconvolve -input OOS-FT-dsMCewsm+orig -progress automask -float - polort A -num_stimts 8 -stim_file 1 /home/yang/doc/fmriref/p4on5off_on_wav.1d -stim_label 1 On -stim_file 2 /home/yang/doc/fmriref/p4on5off_off_wav.1d -stim_label 2 Off -stim_file 3 'mc[0]' -stim_base 2 - stim_label 3 Roll -stim_file 4 'mc[1]' -stim_base 3 -stim_label 4 Pitch -stim_file 5 'mc[2]' -stim_base 4 -stim_label 5 Yaw -stim_file 6 'mc[3]' -stim_base 5 -stim_label 6 ds -stim_file 7 'mc[4]' -stim_base 6 -stim_label 7 dl -stim_file 8 'mc[5]' -stim_base 7 -stim_label 8 dp - num_glt 4 -glt_label 1 On -gltsym 'sym: +On' -glt_label 2 Off -gltsym 'sym: +Off' - glt_label 3 On-Off -gltsym 'sym: +On -Off' -glt_label 4 Off-On -gltsym 'sym: +Off -On' - bucket OOS-FT+mlr -nocout -tout -fout -rout -vout -fitts OOS-FT+fit -errts OOS-FT+ert - xsave -cbucket OOS-FT+cbk -x1d FT" # > Slide 29 fmri Post-Procesing: FT paradigm Intermediate T-map Slide 30 fmri Post-Procesing: FT paradigm Design

retinotopic mapping Language: expressive & receptive speech Expressive speech: Word generation, Object naming, Rhyming Receptive

11 Slide 31 fmri Post-Procesing: FT paradigm Mean + T-map Slide 32 fmri Post-Procesing: FT paradigm GLM Slide 33 Sensorimotor fmri: Typical Tasks Gross motor: Finger-tapping, tongue tapping Fine motor: object manipulation (Mosier) Sensory: Visual field / retinotopic mapping Language: expressive & receptive speech Expressive speech: Word generation, Object naming, Rhyming Receptive speech: Passive Listening, Rhyming Memory: Working memory Other: Swallowing / articulated speech (Mosier) Not yet standardized: ASFNR working on that!

Patient able to undergo MR imaging at 1.5T or 3T (only 3T at IUPUI). AVM patients: specific clips not safe @ 3T.

12 Slide 34 fmri: Choice of Tasks Location Gross Motor Fine Motor Tasks Language Frontal + + /- + Visual Field Working Memory Other Parietal Temporal + / /- Occipital + Object Naming Insular Slide 35 fmri: Patient Selection Intracranial lesion requiring eloquent cortex mapping. Patient able to undergo MR imaging at 1.5T or 3T (only 3T at IUPUI). AVM patients: specific clips not 3T. Stents; not all 3T Body habitus Claustrophobia Able to speak & understand English Able to 6 th grade level. Peds + /- Slide 36 Clinical fmri: Neurosurgical Mapping A fmri (Language rhyming task ) ESC IU Radiology fmri

manipulation right hand: fine motor, sensory tactile, proprioception

13 Slide 37 Case 1: Oligodendroglioma; Bilateral Finger tapping Central sulcus Pre-central gyrus Slide 38 Case 1 Oligodendroglioma; Object manipulation right hand: fine motor, sensory tactile, proprioception Slide 39 Case 1: Oligodendroglioma Language: Word Generation: speech execution Activation

14 Slide 40 Case 1: Oligodendroglioma Language: Naming; semantic language: speech reception and execution Noise Activation Noise Slide 41 Case 1: Oligodendroglioma Language: Rhyming; semantic language Noise Slide 42 Case 2: Finger-tapping

Slide 43 Case 2: Object Manipulation Slide 44 Case 2: Working Memory Slide 45 fmri Brain Mapping Advantages: Non-invasive mapping of eloquent cortex w/ maps co-registered to anatomical images.

15 Slide 43 Case 2: Object Manipulation Slide 44 Case 2: Working Memory Slide 45 fmri Brain Mapping Advantages: Non-invasive mapping of eloquent cortex w/ maps co-registered to anatomical images. In many institutions, this has completely replaced WADA testing. Disadvantages: Not all subjects are candidates: MRI safety, patient must be awake & cooperative, peds. Requires a team with expertise: neuroradiology, neurosurgery, neuropsych., MR physicists, image processing specialists, etc. Indirect measure of neuronal activity.

16 Slide 46 Case 3:WHO Grade II oligoastrocytoma Slide 47 Case 3:WHO Grade II oligoastrocytoma FT TT Naming Rhyming Slide 48 Case 3:WHO Grade II oligoastrocytoma

publication and teaching Slide 51 Case 4: Visual Cortex, publication and

17 Slide 49 Case 4: Visual Cortex, Awake Slide 50 Case 4: Visual Cortex, Awake Note: patient has granted permission for his photos to be used for publication and teaching Slide 51 Case 4: Visual Cortex, Awake Note: patient has granted permission for his photos to be used for publication and teaching

18 Slide 52 Case 4:Oligoastrocytoma < Gr III Left Chkbd Right Chkbd Face Matching Slide 53 Slide 54

19 Slide 55 Slide 56 Fig. 9.3 Slide 57 MR Diffusion Tensor Imaging In liquids In tissues Anisotropy

20 Slide 58 Slide 59 Slide 60 DTI

21 Slide 61 Slide 62 Slide 63 Case 5 - DTI/FA

22 Slide 64 Slide 65 Slide 66 Adapted from: Descoteaux et al DTI: Fiber Tracking

23 Slide 67 DTI at 3T Slide 68 Case 1 DTI Slide 69 Case 2: DTI

24 Slide 70 A C B D Slide 71 Slide 72

25 Slide 73 Case 6: 54 y.o. w/partial complex seizures & speech arrest Slide 74 Case 6: Bilateral Finger-tapping Slide 75 Case 6: Word Generation

26 Slide 76 Case 6: Naming Slide 77 Case 6: DTI Slide 78 Case 6: Perfusion (ASL)

27 Slide 79

functional MRI everything you always wanted to know, but never dared to MD PhD

functional MRI everything you always wanted to know, but never dared to ask @MarionSmits, MD PhD Associate Professor of Neuroradiology Dept. of Radiology, Erasmus MC, Rotterdam (NL) Honorary Consultant