On Mon, 4 Feb 2019 18:01:13 +0000, Christian N. <[log in to unmask]> wrote:
>I assume that the "Size of topology defects" number, which appears in the report sheet after every CAT12 segmentation, indicates the number of uncorrected errors found during surface reconstruction.
The Euler number gives rather a link to the number of defects. However, even for a small number of defects there might be quite large defects that might affect quality of surface reconstruction.
If I am right, do you know if there is any "standard" threshold from which I should start excluding subjects? Would it be wise to include only those subjects with 0 topology errors?
No, this would be quite painful, because this is rather unlikely. You should check data with many defects (large Euler number) and/or larger defect sizes more carefully, but there is no specific threshold or rule of thumb to reject data.