Dear Marcel,
the CAT12 quality measures are based on parameters estimated in white matter (WM) because noise estimates are more stable in WM areas. Therefore, if the segmentation is slightly different between versions, there may be slight changes in the IQR scores. However, values around 2 are fine. Anything below 3 is usually safe to use.
Best,
Christian
On Wed, 29 Mar 2023 08:14:06 +0100, Marcel Daamen <[log in to unmask]> wrote:
>Dear Christian,
>
>I reprocessed a dataset with the most recent CAT12.8.2 release that I previously ran with CAT12.7.
>
>While I had mean IQR scores of about M=1.99 in CAT12.7, with most datasets scoring at least "B-", the IQR scores dropped with the new release, down to M=2.29, with several datasets now scoring in the "C" range.
>
>This drop is clearly driven by worse noise ratings. Did the estimation change between the two releases? I could not find any clues in the Version Information, and I did not notice any changes in the (default) parameter settings of the Segment batch that would provide an obvious explanation.
>
>Best wishes,
>
>Marcel
|