Dear experts,
I'm going to analyse resting state data with MELODIC for the first time,
and I have some doubts.
I've got two groups, patients and healthy controls, and each subject has two rs runs (not in different conditions, simply two runs 4.30 min each).
I've understood I should perform a group-ICA using temporal concatentation, followed by dual regression, but here are my doubts:
- As for the pre-processing, should it be performed separately before MELODIC or is using the pre stats subtab in the MELODIC GUI ok?
- Which is the best way to handle the 2 runs of each subject? simply concatenating them in time in order to have a single 4D dataset for each subject is usually discouraged, but at the same time I'm not sure at which step and how I could somehow average them.
- If I want to denoise my data, by using fsl_regfilt (or FIX), I've read I'd better work on individual sessions, but I don't manage to understand how, or better on which data: should I run MELODIC for each individual session, select noise components, denoise the timeseries and then feed them in a MELODIC group analysis with temporal concatenation (de-selecting all the pre stats options), or is it better to run a MELODIC group analysis, then after dual_regression denoise the timeseries, and finally feed them in an other MELODIC group analysis?
Thank you in advance for your help,
best regards
Stefania E.
|