Dear Edo,
on the HPC I would prefer to call CAT12 from the shell without display. Here you can either use the standalone version of CAT12 (without need of Matlab license) or the standard version:
https://neuro-jena.github.io/enigma-cat12/
The ENIGMA-CAT12 protocol describes the available steps. You can decide with two flags (-ns, -e) whether you call the standalone or standard version.
Standalone version:
/software/CAT12.8.1_MCR/standalone/cat_standalone.sh -m /software/CAT12.8.1_MCR/v93 \
-b /software/CAT12.8.1_MCR/standalone/cat_standalone_segment_enigma.m \
/data/enigma-data/raw/sub*.nii.gz
Standard version:
your_spm12_folder/toolbox/cat12/standalone/cat_standalone.sh -ns -e -m /usr/local/bin/matlab \
-b your_spm12_folder/toolbox/cat12/standalone/cat_standalone_segment_enigma.m \
-s your_spm12_folder /data/enigma-data/raw/sub*.nii.gz
Best,
Christian
On Tue, 19 Jul 2022 13:55:53 +0100, Edoardo P. <[log in to unmask]> wrote:
>Dear all,
>I'm about to analyze many MRI data using the HPC system of my university, so I thought to parallel the segmentation step in order to shorten the time. I know that CAT12 batch has an option to split the task in different jobs, so I was wondering if you recommend to use this option by indicating n processes (based on the cores available on our HPC) or if there is a better way to parallel the whole process.
>
>Thanks a lot!
>Edo
|