Dear FS Team,
We are running recon-all for a large dataset and 5 of the cases keep getting stuck at the exact same spot. It is the point at which the following has just been displayed on the screen:
#@# WM Segmentation Wed Aug 30 17:34:40 EDT 2017
mri_segment brain.mgz wm.seg.mgz
doing initial intensity segmentation... using local statistics to label ambiguous voxels... computing class statistics for intensity windows... WM (107.0): 105.5 +- 6.4 [91.0 --> 125.0] GM (78.0) : 81.0 +- 6.1 [56.0 --> 94.0] setting bottom of white matter range to 87.1 setting top of gray matter range to 93.2 doing initial intensity segmentation... using local statistics to label ambiguous voxels...
For these 5 cases it will sit at this step apparently still running for over a week without any change before I cancel the job. There does not seem to be anything special about these 5 cases either in terms of their header information or visually. We have not found any artifacts in the images except for some minor ringing which is present in many images in the dataset, not just these. The rest of the cases of the dataset all ran to completion and produced all of the files we were expecting, but a subset of them seem to have a related issue where most of the wmparc.mgz labelmap is unsegmented white matter. Only the very edges and the very center are getting designated as something other than this label. The rest of the files in the "mri" directory such as aparc+aseg.mgz and so on are equally as affected by this. Do you know what could be going on here? I would be happy to give you any more information that could be helpful.
Best, Elisabetta and Nate
Elisabetta C. del Re, Ph.D. Assistant Professor of Psychiatry, Department of Psychiatry Harvard Medical School phone 617 9675569 mail elisabetta_delre@hms.harvard.edu