Functional MRI. Jerry Allison, Ph. D. Medical College of Georgia

Transcription

1 Functional MRI Jerry Allison, Ph. D. Medical College of Georgia

2 BOLD Imaging Technique Blood Oxygen Level Dependent contrast can be used to map brain function

3 Right Hand Motor Task

4 Outline fmri BOLD Contrast An fmri Exam Pulse Sequences for fmri fmri Acquisition Parameters fmri Artifacts

5 fmri BOLD Contrast Neuronal events (cerebral activation) O2 consumption increases (+5%) Cerebral blood flow increases (+50%) Oxygen extraction fraction decreases Oxygenation increases in venous blood

6 fmri BOLD Contrast Concentration of paramagnetic deoxyhemoglobin decreases Intravoxel dephasing decreases T2* increases T2* weighted image intensity increases (+ few %)

7 Latency: 5-8 seconds may elapse between neuronal activation and T2* changes

8 Idealized fmri Study Activate Control Activate Control Activate Control T2* Images Voxel Time Course (one voxel)

9 Voxel Time Course Voxel T ime Course Series Image Number Voxel Int ensit y 840

10 fmri Signal from each voxel is characterized by φ :Phase A : Amplitude ω : Frequency T2*

11 T2* has two components 1/T2* = γ B + 1/T2 T2 = nmr spin-spin dephasing γ B = magnetic field inhomogeneity The BOLD phenomenon changes γ B

12 Magnetic Field Inhomogeneity results from Inhomogeneity of the magnet s B0 field Variation in magnetic susceptibility of patient s tissues

13 Magnetic Field Inhomogeneity occurs Near the boundaries of tissues having disparate susceptibility tissue/ air tissue/ bone The sphenoid sinus causes magnetic susceptibility artifacts in EPI images Inferior frontal cortex Inferior lateral temporal cortex

14 Magnetic Field Inhomogeneity also results from The BOLD effect Differences in magnetic susceptibility around a paramagnetic deoxyhemoglobin molecule Changes in magnetic susceptibility around a small blood vessel (capillary, venule, small vein) that has an increased concentration/fraction of oxygenated hemoglobin

15 T2* weighted images are used in fmri to demonstrate changes in magnetic susceptibility associated with the BOLD effect

16 Signal in a T2* weighted BOLD image is affected by Blood volume (CBV) Blood flow (CBF) Arterial hemoglobin concentration Venous hemoglobin concentration Oxygen extraction rate Hematocrit

17 Signal in a T2* weighted BOLD image is thought to be inversely proportional to the number of deoxyhemoglobin molecules in the voxel

18 T2* Contrast is available via: Conventional gradient echo techniques with one RF transmission per phase encoded line of k-space FLASH GRASS FISP

19 T2* Contrast is available via: Single shot EPI techniques with a complete survey of k-space (and subsequent image reconstruction) for each RF transmission EPI - SE techniques EPI - GE techniques Multi-shot techniques (more than one RF transmission per image; more than one echo per RF transmission)

20 EPI Images EPI - GE Technique (T2* weighted)

21 To Map Brain Function Acquire T2* weighted images during a brain task Images will have slightly higher intensity in active brain regions Acquire T2* weighted images during a control state with the brain task suspended Statistically subtract control images from task images to map areas of brain activation associated with the task

22 An fmri Exam involves: Brain Activation Paradigm Task presentation systems Patient response monitoring Image acquisition Synchronization Data processing

23 A brain activation paradigm Control task Activation task: a task designed to produce brain activation (e.g. motor, sensory, language, memory) Try to avoid Habituation Learning Inattention

24 Event Related fmri fmri need not be constrained by task on/ task off block designs By measuring BOLD response to brief stimuli (typically presented at irregular intervals), it is possible to characterize the hemodynamic response function

25

26 Noun Verb Task

27 Right Hand Motor Task

28 Left Hand Motor Task

29 Auditory Task

30 Task presentation systems Audio system w/ Input for external sound sources such as computer audio, VCR, stereo, microphone Attenuate gradient noise while enabling communication Non-pneumatic audio offers improved quality Electrical stimulation A dark room and dark magnet bore

31 Task presentation systems Visual presentation Slide projector LCD panel & overhead projector & rear screen projection Large screen LCD MRI projection systems Electronic goggles MRI compatible corrective lenses

32 fmri Projection System

33 fmri I/O Devices

34 Response monitoring In order to document whether the patient is doing the task Key pad Joystick Track ball

35 MRI Control Room

36 Synchronization Task presentation and patient response monitoring should be in synchrony with the acquisition of T2* weighted MRI images Ideally, all of this apparatus should be controlled by one host.

37 Signal conditioning Wires and tubes that pass in/out of the exam room must pass through RF filters or RF waveguides

38 Penetration Panel

39 Image acquisition T2* weighted images are acquired while the patient alternates between periods having an activating task and periods of a control state.

40 Motor task Begin imaging Control state Rest for 30 seconds Activating task Finger tapping w/ Rt. hand for 30 seconds Alternate these activities for 6 minutes

41 Motor task Acquire 120 T2* weighted MRI image sets Each set: Transverse oblique 32 Slices Primary motor cortex to cerebellum

42 Motor task Thus for each brain voxel, we have temporal data (the voxel time course) having 120 data points (a sample every three seconds) There can be hundreds of thousands of brain voxels

43 Motor task 3840 total EPI images Some scanners limit the number of images in one study (512, 2048, etc.)

44 Sagittal Localizer

45 We use one of two pulse sequences ep2d_fid_66b1190_62.ekc ep2d_fid_60b2080_62_64.ekc

46 ep2d_fid_66b1190_62.ekc Gradient refocused single shot EPI technique Uses optional hybrid gradient overdrive amplifiers Matrix: 128x128

47 ep2d_fid_66b1190_62.ekc TE: 66 msec TR: 3 sec We acquire 22 slices every 5 seconds

48 ep2d_fid_66b1190_62.ekc 22 slices 2.0 mm skip 1.0 ascending Orientation: Transverse oblique: T --> C, - 15 o

49 ep2d_fid_66b1190_62.ekc Phase encode: A P (for symmetry) FOV: 230x230 mm: Voxels: 1.8 x 1.8 x 3.0 mm

50 ep2d_fid_66b1190_62.ekc

51 ep2d_fid_60b2080_62_64.ekc Enables up to 64 slices in one file We do 32 slices every 3 seconds 64 x 64 matrix All slices are written into x 512 image (mosaic mode) Reduces I/O time necessary to write data to disk

52 ep2d_fid_60b2080_62_64.ekc

53 GRASS fmri Technique TR: 70 msec TE: 40 msec Flip angle: 40 o Matrix: 128 x 256 Slice thickness: 6 mm Acquisition: 9.4 sec per image

54 Computing options MRI scanner software AFNI STIMULATE SPM Other approaches Brain Voyager MEDx

55 Data processing may include the following Correction for differences in time of acquisition across slices Examine the data Motion detection Plot center-of-mass time course for each slice Stimulate View Cine loop of T2* weighted images As native signal intensity As a difference image (a. la. DSA) fmri activation in periphery of brain is an indication of motion

56 Data processing may include the following Motion correction Discard images showing obvious motion Image re-registration to correct for rigid body motion 2D (3 parameters) 3D (6 parameters)

57 Data processing may include the following Baseline flattening 0th order 1st order 2nd order fmri image calculation Clustering of activated voxels

58 Data processing may include the following Image fusion With T2* weighted images With T1 weighted images Multi Planar Reconstruction of fused images Volume rendering of fused images

59 Motion: Peripheral Brain Activation

60 3D Image Registration

61 Image numbe r Re f erence Value Reference Curve Ref erence Curve Series

62 Cross Correlation

63 fmri on T2* weighted EPI images

64 fmri on T1 weighted images

65 fmri on Orthogonal T1 weighted images

66 EPI (single shot) Sequence Problems Nyquist ghosts: N/2 ghosts caused by odd and even echo asymmetries Can apply a correction by measuring the asymmetry with a phase reference FID with the phase encode gradient switched off just prior to each single shot image

67 GHOSTS!

68 Artifacts in fmri images Subject motion Bulk Physiologic Cardiac Respiratory Acquisition time (long is bad: like mammography)

69 Artifacts in fmri images ROI (Are there magnetic susceptibility artifacts in the ROI) Geometric distortion Poor shimming results in additional geometric distortion (particularly in gradient recalled EPI images)

70 Susceptibility Artifact in Orthogonal T2* images

71 Phantom Susceptibility in a T2* Weighted Sequence

72 Physiologic Motion Bulk motion Settling

73 Physiologic Motion Respiration Susceptibility changes caused by movement of chest during respiration (phase changes of 2o 6o at 40 msec) Slow varying Affects inferior images Proportional to B0 Proportional to TE

74 Physiologic Motion Cardiac pulsations Non-rigid body movement of brain parenchyma caused by cardiac pulsations (Brain changes shape) Image dependent brain deformation in EPI Image dependent motion artifact in FLASH Can produce artifacts as large as a few % of BOLD contrast Occurs near CSF (Brainstem, Cerebellum)

75 Motion of as little as 0.1 pixels can seriously degrade fmri images because the MRI signal variation from pixel to pixel is larger than the BOLD effect that is being measured.

76 Motion prevention An Ounce of Prevention is Worth A Pound of Cure Padding Expandable foam (Alpha Cradle) Vacuum bags Hammock Bite bar Contour masks

77 Motion prevention Real time Pressure sensors Infrared systems

78 fmri Head Fixation