Hey all,

I'm running into a non-reproducible but persistent problem with CONN (v17.e) during the step where the spatially preprocessed functional data is read and stored in conn_*/data/DATA_Subject*_Session*.matc files.  CONN seems to be occasionally skipping a slice for a given frame, setting all voxels to 0 instead (see attached .nii file which shows 1 frame out of 500 - it is the only frame with a "missing" slice).  The result is that after denoising - i.e. in the conn_*/results/preprocessing/DATA_Subject*_Condition*.matc files - all timepoints of the voxels in the affected slice have wildly distorted values with positive and negative intensities far greater in magnitude than the rest of the brain.

If the Setup step is redone for that subject, the same frame/slice is not affected in the same way.  Re: the prevalence of this issue: in 25 subjects each with 500 frames, 3 subjects had issues with 1, 4, and 4 frames, respectively.  In one instance three consecutive slices in a given frame were affected, which makes me think this might be a "resource busy" read/write error.

My current solution is re-running the Setup step for subjects and checking the ~/preprocessing/NORMS* files until everything looks okay, but I'm looking for something a little less tedious.  Thoughts?  Suggestions?  Thanks!
Jon