Hi! I am using fmriprep-docker.
I have a big dataset 7T. So, I am encountering many errors which seem to be related to the limitation of cache memory. In fact, fMRIprep works well when I conduct it with a subset of the data. So, I wonder whether I can conduct fMRIprep with a subset of the data first and then reuse the results to the other remained data.
Specifically, suppose that the data is composed of 2 repetitive sessions with 2 tasks and T1w on each session. Basic structure would be like this:
- sub-01
- ses-01
- anat
- sub-01_ses-01_T1w.json
- sub-01_ses-01_T1w.nii.gz
- func
- sub-01_ses-01_task-A_run-1_bold.json
- sub-01_ses-01_task-A_run-1_bold.nii.gz
- sub-01_ses-01_task-A_run-2_bold.json
- sub-01_ses-01_task-A_run-2_bold.nii.gz
- sub-01_ses-01_task-B_run-1_bold.json
- sub-01_ses-01_task-B_run-1_bold.nii.gz
- sub-01_ses-01_task-B_run-2_bold.json
- sub-01_ses-01_task-B_run-2_bold.nii.gz
- ses-02
- anat
- sub-01_ses-02_T1w.json
- sub-01_ses-02_T1w.nii.gz
- func
- sub-01_ses-02_task-A_run-1_bold.json
- sub-01_ses-02_task-A_run-1_bold.nii.gz
- sub-01_ses-02_task-A_run-2_bold.json
- sub-01_ses-02_task-A_run-2_bold.nii.gz
- sub-01_ses-02_task-B_run-1_bold.json
- sub-01_ses-02_task-B_run-1_bold.nii.gz
- sub-01_ses-02_task-B_run-2_bold.json
- sub-01_ses-02_task-B_run-2_bold.nii.gz
In this case, what is the most memory-light pipeline to conduct fMRIprep to all of the data?
My conjecture is that first, I implement fMRIprep only to a single task (e.g., task-A) of the single run (e.g., run-1), and second reuse the results of the first step to the other remained bold signals (e.g., task-A, run-2 & task-B, run-1 & task-B, run-2). If my conjecture is right, can you give me any advice to implement it? or, is there any more efficient way to achieve my goal?
Many thanks in advance!!