Greetings FSLers,
I am encountering a rather unsettling result on running Temporal concatenation ICA analyses. I am getting time-courses that are mostly flat until we get to the last quadrant of the X axis where I am getting crowded spikes, meanwhile my frequency power-spectra are all over the scale (they look more like time-course plots to me).
This only happens with temporal concatenation ICA, if I run single session ICA the time-courses and powerspectra are acceptable, and it only happens when I run the analysis on FSL 5.0.9 with MELODIC 3.14 under Linux OS (Ubuntu 14, 12 and Debian 8). I downloaded FSL from the Neurodebian repository every time, using the fsl-5.0-complete package. I have FSL 5.0.8, MELODIC 3.14 running on Mac OS X Yosemite and I've never encountered such an issue with it. I ran the same datasets on the Mac environment as I did on the previously mentioned Linux distros, to make sure it wasn't a problem with the data. I also checked the filtered_func_data.nii.gz files inside the .ica directories in FSLview with the timeseries option, and they didn't show anything similar to the ones shown on the temporal concatenation ICA report, so I don't think it's an issue with the preprocessing pipeline, but rather something wrong with the Multisession Temporal concatenation ICA step itself.
Kindly advise, is this a bug ? is there something I can do to fix it ? Am I doing something wrong with the installation ?
Thanks in advance,
Radwan