ISC (inter-subject correlation) is a medical imaging method that gives you the segments of the brain that behave similarly when under similar stimuli.
I am essentially calculating a voxelwise correlation of a bunch of 5 MB files across 29 folders and I`m running into performance issues. I tried the Dagger module, but It didn’t go well.
The code is in this repo if you have some suggestions https://github.com/AShedko/fMRI_ISC_classify/tree/master/Jthing
Although it`s useless without the data and there is 150 gigs of it