CAT26 T1w grey matter (longitudinal) segmentation and expert GUI setting/tracing issues

[Sarah_Tischer]

Hi CAT users and other experts,

I have multiple issues with longitudinal preprocessing of of T1w scans in CAT26.0.rc3/4 (in SPM25.01.02 in Matlab 2025a) with the standard and expert GUI.

Problem A:

My pediatric (8-12 y/o) (longitudinal) segmentations have big chunks of sinuses wrongly classified as GM, although I use CAT’s pediatric tissue probability maps (TPM):

I have tried to modify several settings like:

1. Strength of Final Clean Up (cleanupstr): strong (0.75), heavy (1.00)
   → seems to have a positive impact
2. Modify Affine Scaling (affmod): -4-10%
   → -4-6% seems to have a positive impact (less sinus inclusion), but it is first applied to the average image, and then again to the individual image, so the brain/skull TPM overlay is too small and cuts off GM in the edges (mainly temporal inferior and frontal lobes)
   
   
3. Skull-Stripping (gcutstr): APRG approach V2 tighter (2.9), APRG approach (force skull-stripping, 12)
   → unclear since usage is unclear (see problem B)
4. SPM processing accuracy: very high (very slow)
   → seems to have a negative impact (includes more sinus) on dorsal parts
5. Strength of Blood-vessel correction: allways (1.00)
   → worsened segmentation significantly

Fiddeling with the settings had some positive effects, but the main negative effects are missing GM in the inferior tempral lobe(s).

Problem B:

I am generally unsure if all settings via the (expert) GUI are actually applied because there is no transparend documentation:

1. For skull-stripping, the catreport indicates the default APRG approach (2.0) and APRG approach V2 tighter (2.9) correctly when used, but when choosing APRG approach (force skull-stripping, 12) the catreport only indicates the default APRG(2.00). How can I verify the setting was really used?
2. When I select the CAT pediatric tissue probability map (TPM_11.5.nii) it is displayed as used on the quality report (catreport), but the xml file only indicates the default:
   </filedata>
<parameter>
<vbm>
<opts>
<tpm>
<item>/Users/sarah.tischer/Documents/MATLAB/spm/tpm/TPM.nii</item>

Questions:

1. Should I expect to find a trace of my settings in the .xml files?
2. How can I verify what settings were applied, especially when the input is not printed in the catreport and/or xml file?
3. Has anyone ideas about how to get more precise grey matter segmentations, what (other) parameters to change?

Please let me know if mor detail is needed here.

Thanks so much,

Sarah