I am interested in applying RIPTiDe to fMRI data for which I also have physiological recordings (ECG, EDA, and respiration- not NIRS), but am unsure about which settings to use when calling rapidtide.
Rapidtide only takes in one regressor at a time, right? Should I then run it twice- once with ECG and once with respiration?
In terms of workflow parameters, I assume that I would want to use the associated --filterband for each (i.e., resp for respiration and cardiac for ECG), but should I do the standard preprocessing to these signals (e.g., downsampling, peak detection, convolution with RRF/CRF) or does rapidtide expect the raw data?
If my goal is to retain the voxel-wise nuisance regressors for a later GLM, would it be sufficient to just use the --denoising macro and keep everything else at its default?
@tsalo - Hey Taylor. Working to implement Rapidtide as well. I am using BIDS .tsv.gz physio-respiratory and .tsv.gz physio-cardiac files. Curious to hear what implementation you settled on and any helpful documentation that you come across? I noticed that happy tries to estimate a better cardiac waveform, and so am esp curious how others have approached this task.
Hi @mdemi. Unfortunately, I never ended up combining the physio with rapidtide. I ended up using each separately to compare denoising strategies. I’d love to know how best to do it though.
if you’re still looking for solution, I have made a toolbox that has similar functionality to rapidtide but focuses more on the inclusion of physiological readings like PetCO2. Happy to think along with your analysis.