Post on 25-Mar-2020
Preprocessing of fMRI data
Preprocessing of fMRI data
Pierre Bellec
CRIUGM, DIRO, UdM
Preprocessing of fMRI data
Flowchart of the NIAK fMRI preprocessing pipeline
fMRI run 1
fMRI run N
CIVETNUC, segmentation,spatial normalization
Motionestimation
coregistration
- median fMRI
regressconfounds
run 1
regress confounds
run N
resamplingrun 1
resamplingrun N
individualdatasets
slice timingrun 1
slice timingrun N
scrubbingrun 1
scrubbingrun N
CORSICArun 1
CORSICArun N
smoothingrun 1
smoothingrun N
defaultoff
Preprocessing of fMRI data
Slice timing correction
interleaved mode
TR ∆t2∆t1
fMRI time series
∆t1∆t2
1
2
3
4
5
6
7
8
9
temporal interpolation to a single reference timefor each volume (cubic spline interpolation)
Courtesy of Dr M. Pelegrini-Issac.
Preprocessing of fMRI data
Motion estimation: within-run
within-run motion parameters :3 rotations
3 translationsfor each volume
volume of reference (median)
time
time
Preprocessing of fMRI data
Motion estimation: between-run / within-session
between-runmotion parameters
between-runmotion parameters
RUN 1
RUN n
RUN N
time
Preprocessing of fMRI data
Motion estimation: between sessions
between-runmotion parameters
between-runmotion parameters
RUN 1
RUN n
RUN N
time
between-runmotion parameters
between-runmotion parameters
RUN 1
RUN n
RUN N
time
between-runmotion parameters
between-runmotion parameters
RUN 1
RUN n
RUN N
time
between-sessionmotion parameters
between-sessionmotion parameters
SESSION 1 SESSION k SESSION K
Estimation of between-run (between-session) rigid-body motion.
Preprocessing of fMRI data
T1 processing: linear coregistration
Preprocessing of fMRI data
T1 processing: non-linear coregistration
Preprocessing of fMRI data
T1 processing: linear template
Linear ICBM template (average of 152 subjects)
Preprocessing of fMRI data
T1 processing: linear coregistration
Individual structural scan (linear coregistration)
Preprocessing of fMRI data
T1 processing: non-linear coregistration
Individual structural scan (non-linear coregistration)
Preprocessing of fMRI data
T1 processing: nonlinear template
Symmetric non-linear ICBM template (average of 152 subjects)release 2009a.
http://www.bic.mni.mcgill.ca/ServicesAtlases/ICBM152NLin2009
Preprocessing of fMRI data
T1 processing: group average
Average of 17 subjects (non-linear coregistration)
Preprocessing of fMRI data
T1 processing: Flowchart of the CIVET pipeline
non−uniformitycorrections
brain masking
linearregistration
non−uniformitycorrections
brain masking
non−linearregistration
native MRI stereotaxic MRI brain masking
classification
Flowchart of the T1 preprocessing.
Preprocessing of fMRI data
T1 processing: main outputs
raw T1 volume
non-uniformity corrected volume
brain mask
linear MNI152 template
non-linear MNI152 template
linear (lsq9) coregistered volume
non-linear coregistered volume
The main outputs of the T1 processing pipeline.
Preprocessing of fMRI data
Coregistration between the T1 and fMRI volumesC
orr
ect
to z
ero
mean, sm
ooth
, ere
sam
ple
Corre
ct to
zero
mean, sm
ooth
iteration 1 : large smoothing
iteration 5 : small smoothing
5 iterations progressive fit : LSQ6 coregistration - MINCTRACC (mutual information)
iteration 1 : large smoothing
iteration 5 : small smoothing
T1 volume & brain mask fMRI volume & brain mask
smooth : 8,4,8,4,3step : 4,4,4,2,1simplex : 8,4,2,2,1
smooth : 8,4,8,4,3step : 4,4,4,2,1simplex : 8,4,2,2,1
Preprocessing of fMRI data
Spatial resampling
native functionalspace
stereotaxic space -individual volume -
non-linear transform -
stereotaxic space - average of 40 subjects
The transformations to correct for rigid-body motion during the fMRI acquisition and
the transformation to match the T1 image and then (non-linearly) coregister into
stereotaxic space are all combined, and a single step of spatial resampling is applied.
Preprocessing of fMRI data
Scrubbing: frame displacement
Frame displacement is the sum of absolute displacements in translation and rotation
motion parameters. For each frame with excessive FD (here FD> 0.5), four frames are
suppressed (the target one + one before + two after, marked with red stars on the
figure). The original method was proposed by Power et al. Neuroimage 2012. Note
that, unlike the original method, only FD is used in NIAK (and not DVARS).
Preprocessing of fMRI data
Scrubbing: example of impact on the DMN
See Power et al. Neuroimage 2012 for more infos.
Preprocessing of fMRI data
Regress confounds: model
Preprocessing of fMRI data
CORSICA: sources of structured noise
Cardiac fluctuationsDagli et al. Neuroimage (1999)
Respiratory fluctuationsRaj et al. Phys. Med. Biol. (1999)
Residual motionMcKeown et al. HBM (1998)
1/f acquisition noisespace-time filteringslice-timing correctionmotionmotion correction
Many sources of space-time correlated noise fluctuations are corrupting fMRI time-series:
Preprocessing of fMRI data
CORSICA: independent component analysisgro
up 1
gro
up 0
gro
up 2
spatially independent components analysisPerlbarg et al. Magnetic Resonance Imaging, 2007, 25: 35-46.
Preprocessing of fMRI data
CORSICA: selection of “noise” components
Flowchart of the CORSICA algorithm for correcting structurednoise in fMRI
Perlbarg et al. Magnetic Resonance Imaging, 2007, 25: 35-46.
Preprocessing of fMRI data
CORSICA: effect map
x = 0 y = 0 z = -18
0 0.4
Relative variance of estimated structured noise using CORSICA. Averageon 40 subjects, 5 tasks per subject.
P. Bellec, V. Perlbarg and A. C. Evans, Magnetic Resonance Imaging, 2009, pp. 1382-1396..
Preprocessing of fMRI data
Spatial smoothing
native resolution
smoothed imageisotropice Gaussian kernel - 6 mm FWHM
Preprocessing of fMRI data
Quality control and assessment
It is recommended to check the quality of the following steps:
Coregistration of the individual T1 image and the braintemplate (stereotaxic space).
Coregistration of the individual T1 image and the individualaverage BOLD volume.
Ammount of motion.
Complete guidelines: https://github.com/SIMEXP/niak_manual/raw/master/qc_manual_v1.0/qc_manual_niak.pdf
Preprocessing of fMRI data
Quality control and assessment: example
Preprocessing of fMRI data
Useful links
1 The download page, with this pdf presentation, NIAKreleases and the demo datasethttp://www.nitrc.org/frs/?group_id=411
2 The NIAK online user’s guidehttp://www.nitrc.org/plugins/mwiki/index.php/niak:MainPage
3 The NIAK project page and developer’s guidehttps://github.com/SIMEXP/niak/wiki
4 The PSOM project pagehttp://psom.simexp-lab.org/
Preprocessing of fMRI data
License
The NIAK project is under an MIT opensource license
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated docu-mentation files (the ”Software”), to deal in the Software without restriction, including without limitation the rightsto use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit personsto whom the Software is furnished to do so, subject to the following conditions:The above copyright notice and this permission notice shall be included in all copies or substantial portions of theSoftware.THE SOFTWARE IS PROVIDED ”AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, IN-CLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULARPURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BELIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THEUSE OR OTHER DEALINGS IN THE SOFTWARE.