Rapidtide Memory Usage

Software Support
1

Hi all,

I am currently interested in using rapidtide to calculate the latency from stroke patients’ rsf-MRI. However, due to my current location, I am forced to do it from my personal computer which has a RAM memory of 8 gb (3.6G for WSL2). Would it still be possible to run rapidtide? I tried it several times using singleproc options and a brain mask, but it seems like the processors always overloads and is unable to run my analyses. I am relatively new with these analyses so your help would be much appreciated. Below I’ve provided some screenshots regarding the issue:

1
11914×563 301 KB

2

I’ve been working on a BIDS App using rapidtide (fMRIPost-rapidtide) and I’ve noticed the same issue. @bbfrederick made a few recommendations in Issues · nipreps/fmripost-rapidtide · GitHub

  • Use --spcalculation
  • Split the regressor estimation and denoising steps into separate steps. Run rapidtide with --noglm first, then run retroglm on the outputs.
  • It might work better if you process the data in boldref/native space/resolution, then warp the results to standard space.