Preprocessing of fMRI data - Amazon S3 · Preprocessing of fMRI data Scrubbing: frame displacement Frame displacement is the sum of absolute displacements in translation and rotation

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.