 |
 |
Hi FSL experts, I am learning FSL, after using AFNI for quite some time, and I am trying to compare processing streams and what each package is doing differently. There is one major difference that I have noticed. In AFNI, we usually used the lower-level "percent signal change" file (p04_scales from afni_proc.py) and/or the statistics file (output from 3ddeconvolve) as input for second-level / group level analyses. However, FSL uses the cope (parameter estimates from the contrast) as input for second-level/group-level analyses. For example: A) PPI seed time course in AFNI is extracted from the percent signal change file. In FSL, the seed time course is extracted from the lower level cope files (usually filtered_func_data). B) For ROI analyses in AFNI, the single-subject values for the ROI mask are extracted from the stats file of the relevant contrast that was generated from 3ddeconvolve. As far as I know, FSL users extract the single-subject values from lower level cope files (filtered_func data), not from the statistics files. C) For group-level analyses (for example, group t-test) the input for AFNI 3dttest is the lower-level stats file of the relevant contrast. In FSL, the input for a higher-level FEAT analysis is the lower level cope file for each subject, not the statistics file. Perhaps these differences are not package specific, and maybe it's just the way I have learned what to do in each package. However, I am wondering if anyone has a conceptual explanation of (a) how cope files differ from percent signal change in FSL (b) whether you would get different results using a stats file vs. a cope file as input for higher-level group analyses and (c) whether FSL recommends only using cope files for higher level analyses. Thanks! Michelle