Basic terminology for different scales

1
Basic terminology for different scales

• session
• run
• volume
• slice
• voxel

2
Image Type 1 Scout (localizer)

• First acquire quick scout images
• One each
• Axial
• Sagittal
• Coronal
• Allows for headpositioningslice positioning
• Ignore afterwards

3
Image Type 2 High-res Volume

• MPRAGE T1-weighted sequence
• Typical dimensions
• 256 x 256 pixels/slice
• 128 slices
• 1mm cubic voxels
• (example axial, but we do sagittal)
• Acquisition time
• 6-10 minutes

4
Image Type 3 Functional

• EPI T2 weighted sequence
• Typical dimensions
• 64 x 64 pixels/slice
• 10 - 30 slices
• 3x3x(3-6) mm voxels
• Acquisition time
• 1.5 - 3 sec
• This speed, plus theBOLD response,
  enablesfunctional MRI

5
Compare MPRAGE vs. EPI

• Differences are
• Speed minutes vs. seconds
• Contrast T1 vs. T2
• Resolution high vs. low
• high-res MPRAGE in 3D vs. low-res EPI in 3D

6
EPI speed enables fMRI

• 1 slice from EPI
• New image every 2 seconds
• Look at small signal changes
• Map changes to experimental events

7
Functional Time Series

• Look at one voxel of EPI images, over time

Time (TRs) --gt
Some changes due to task manipulations (colored
bars),others are noise masking the desired
effect
8
BREAK for DESIGN
9
fMRI Data Preprocessing

• Importing viewing different types of MR data
• Functional (EPI)
• Volume (MPRAGE)
• Other anatomical (In-plane T2 SE, T1, scout,
  etc.)
• Preprocessing functional data
• 3D motion correction
• Slice timing correction
• Spatial filtering
• Temporal filtering
• Less-used corrections (EPI distortion, mean
  intensity)
• Spatial Normalization
• Aligning data to common reference brain (e.g.,
  Talairach)

10
Preprocessing to Diminish Noise

• 3D motion correction
• Slice scan timing correction
• Spatial filtering
• Temporal filtering
• Less-used corrections
• EPI distortion
• mean intensity

11
3D motion correction the Problem

• Subjects move, like it or not
• EPI image also moves from B0 shift
• Along phase encode
• (L-gtR or A-gtP)
• If voxel at high-contrast edge, then time
  series will show changes just from motion

12
3D Motion Correction the Solution

• Computer tries to align all images over time
• Typically, align each image to first image in run

13
3D Motion Correction the Solution

• Use 6 parameters of rigid-body motion
• Shift X, Y, Z (in mm)
• Rotate X, Y, Z (pitch, roll, yaw in degrees)

14
Preprocessing to Diminish Noise

• 3D motion correction
• Slice scan timing correction
• Spatial filtering
• Temporal filtering
• Less-used corrections
• EPI distortion
• mean intensity

15
Slice timing correction the Problem

• Slices of an EPI image acquired separately
• Therefore, information from different slices come
  from varying points in time after task events
• Different orders of slice acquisition
• ascending vs. descending
• sequential vs. interleaved

16
Slice scan timing correction the Solution

• Correction shifts each voxel's time series so
  that all voxels in a given volume "appear" to
  have been captured at exactly the same time

The slices of one functional volume are shifted
in time.
This is correctedby sinc (or linear)
interpolation in time.
17
Preprocessing to Diminish Noise

• 3D motion correction
• Slice scan timing correction
• Spatial filtering
• Temporal filtering
• Less-used corrections
• EPI distortion
• mean intensity

18
Spatial filtering

• Gaussian spatial filtering blurring
• gt SNR if spatial filter same size as activation
• Don't use TOO much!

19
Spatial Filtering

• Results with different smoothing FWHM values (in
  mm)

20
Preprocessing to Diminish Noise

• 3D motion correction
• Slice scan timing correction
• Spatial filtering
• Temporal filtering
• Less-used corrections
• EPI distortion
• mean intensity

21
Temporal Filtering the Problem

• The time series from each voxel contains
• low frequency drifts
• high frequency noise
• caused by
• scanner
• physiology
• Both can hide your effect

22
Temporal Filtering the Solution

• FFT, then remove frequencies high, low or both,
  without removing the frequencies of interest
• Lowpass temporal blurring
• Be careful don't remove your effect!
• Highpass detrending
• Bandpass both

23
Preprocessing to Diminish Noise

• 3D motion correction
• Slice scan timing correction
• Spatial filtering
• Temporal filtering
• Less-used corrections
• EPI distortion unwarping
• Helps EPI align correctly with structural scans
• Like motion correction, different parameters
• Mean intensity adjustment
• Not recommended anymore by anyone!

24
Effect of Preprocessing
before