Hi Juan,

Just a quick follow-up:

Regarding using FSL realignment parameters, if you simply enter directly (without changing the extension name) the MCFLIRT outputfile.par files into CONN as a first-level covariate named "realignment", CONN should interpret the order and magnitude of those FSL-generated motion parameters correctly, so I am curious as to why this did not work in your dataset. 

Just for reference, the four different motion parameters accepted in CONN are: a) rp_*.txt files in SPM-format (3 motion mm  + 3 rotation radians); b) *.par files in FSL-format (3 rotation radians + 3 motion mm); c) *.deg.txt files in HCP-format (3 motion mm + 3 rotation degrees); and d) *.siemens.txt files (in-scanner motion correction format; 3 resorted motion mm + 3 rotation degrees)

By default CONN will assume SPM-format (a) unless the file extension matches any of the alternative cases (b-d above), so I imagine perhaps the issue in your dataset was that the FSL-generated parameters had been perhaps renamed to a different file extension? Please let me know

And regarding the other issue (index exceeds matrix dimensions error), my impression is that this error was related to the same mis-interpretation of the motion parameters as well. CONN was trying to use the art-generated mean functional volume (robust mean functional image computed disregarding any outlier scans) for normalization&segmentation, but the problem was perhaps that this image had not been created correctly because all scans had been flag as outliers, hence the strange index-exceeds-dimensions error message. Summarizing, if you prefer to use the mean functional image for normalization instead of the first functional volume, which should be generally a bit more robust, please try again after fixing the motion-parameters issue above and (I believe) the second issue should resolve itself. 

Hope this helps
Alfonso
Originally posted by Juan Dominguez: