Looks like the thread got split up; original posts copied below:
---
Hello FSL team,
I'm analyzing some fMRI data that have been denoised using MELODIC ICA (specifically, resting-state data from the Human Connectome Project denoised via ICA-FIX), and I'd like to determine the percent of variance explained/removed from a particular ROI. The ICA denoising was run on whole-brain data, so there is already a report indicating the percent of explained variance corresponding to each component across the entire brain. Is there a way of determining the analogous values for a masked region?
Thank you very much!
-Ely
---
Hi Ely,
Do you care about each component separately or is all of them combined
more what you are after? Also, do you want a number for each voxel in the
ROI or all of them?
Peace,
Matt.
---
Hi Matt,
Hmm, good questions. The contribution of each component and the voxelwise values would also be of interest, but getting the total variance removed from the whole ROI is the most important thing. Presumably I could get the voxelwise values by repeating the calculations using single-voxel masks, if needed.
Thanks!
-Ely
|