Scripts here are related to fMRI dataset 'studyforrest_movie_denoised'(https://openneuro.org/datasets/ds001769).
In the 'studyforrest_movie_denoised', we denoised the studyforrest audio-visual movie fMRI data(https://openneuro.org/datasets/ds000113) following a four-step procedure including 1.preprocessing, 2.ICA decomposition, 3.IC classification and 4.artifacts removal.
-
Preprocessing
Discription: performed using FEAT in FSL version 6.00 (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/).
Code: /preprocessing.sh -
ICA decomposition
Discription: performed with a probabilistic ICA algorithm implemented in the FSL’s MELODIC version 3.15 (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/MELODIC).
Code: /MELODIC_ICA.sh -
IC classification
Discription: Classification of ICs was done manually using Melview (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Melview). -
Artifacts removal
Discription: performed with FSL’s MELODIC version 3.15
Code: /artifact-IC_removal.sh
Additionally, the technical quality of the datasets was validated in two aspects, temporal signal-to-noise ratio (tSNR) and inter-subject correlation (ISC)
-
Register fMRI volume data on 'fsaverage' surface template using FreeSurfer version 6.0.0 (https://surfer.nmr.mgh.harvard.edu)
Code: /validation_analysis/preprocessing_fsaverage.sh -
Calculate tSNR and ISC for pre- and post-denoising fMRI data
Code: /validation_analysis/tSNR.py
/validation_analysis/ISC.py -
Calculate the cohen's d effect size for tSNR and ISC between pre- and post-denoising fMRI data
Code: /validation_analysis/cohen_d.py -
Generate visualization of the cohen's d of tSNR and ISC
Code: /validation_analysis/results_visualization.py