You don't necessarily have to exclude them as you can use weighted random effects (it will de-weight the noisy scans). All the same, it is a good idea to try to track down why a few are noisier. The prime suspect is motion, so check the motion plots. There might be scanner spikes as well (try spikedet-sess). In rare instances, the thermal/white noise may have increased in your scanner on those acqs -- this is a hard thing to diagnose though.
doug
Christian Battista wrote:
Hi freesurfers,
I am using FSFAST on a group of 29 subjects...having run isxconcat-sess I am now doing the final quality control (Volume) step outlined in the FSFAST tutorial, which is to inspect the SNR maps. Inspecting the signal to noise ratio maps I am seeing 3 outliers (see attached), suggesting there is something strange going on with 3 participants' trials. I assume this means I should be excluding these participants from the analysis...but I am also wondering what the cause of the outliers might be, and whether it might be the result of a failed step earlier in the analysis. Is there anywhere else in the processing stream that I would be able to notice something is wrong?
Take it easy, -Christian
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer