FAQ

Q: Where do I find the average surface and related surface models for statistical analyses?

A: Look in Vertex-Based Statistical Analyses or get it from CIVET_2.0.tar.gz for CIVET-2.0.0 and CIVET-2.1.0. Note that this is the model to use with the marching-cubes algorithm starting at CIVET-2.0.0.

Q: CIVET fails. Can I restart it?

A: If CIVET fails during extraction of the white surface (e.g. stage extract_white_surface_right), there is nothing to do. This is caused by self-intersections in the resampled surface after surface registration and surface extraction cannot continue. There is no need to restart the job as it will continue to fail. We are working on fixing this.

A: If CIVET fails during the PVE stage, it’s likely that the brain mask is empty. Inspect your input T1w image for a possible problem. Don’t be surprised if the input T1w image is empty or incomplete.

A: If CIVET fails at some other stage, you may try to restart it.

Q: Can CIVET produce high-resolution surfaces?

A: The high-resolution surfaces are not quite ready in CIVET. Yes, the pipeline can produce them, but they are basically a sub-division of the low-resolution surfaces, with incomplete convergence. The current approach is to extract the low-resolution white and gray surfaces, sub-divide, then converge. That’s not really optimal. The new version will extract the high-resolution white surface directly from the marching-cubes initial white surface, then converge it. The gray surface expansion will start from the high-resolution white surface. It’s a bit slower than starting from the low-resolution surface, but surfaces are fully converged and more accurate. Note that high-resolution surfaces have been momentarily been disabled in CBRAIN.

Q: When do I need to use the switch -mask-blood-vessels?

A: Major blood vessels usually appear as high intensity voxels in the T1w image. If this is the case, they will be erroneously misclassified as white matter voxels and these erroneous voxels may interfere with the extraction of the white matter surface. If errors in the morphology of the white matter surface are detected and are due to the presence of blood vessels, then re-running CIVET with -mask-blood-vessels often fixes the problem. The blind usage of this option on all scans is discouraged since thresholding is used to detect blood vessels. If no blood vessels are present, then thresholding will occur on true white matter voxels, consequently causing missing bits in the white matter surfaces. Only use this option when needed.


Previous Section: References
CIVET Home