ASHS preprocessing and config

[Coco Newton]

Hi Paul and other ASHS users,

I have a couple of generic ASHS usage questions that I can't find answers for after digging around here, in the ASHS scripts or the cited papers. I'm trying to troubleshoot poor ASHS-T2 performance on my 7T data (T1W 0.7 iso, T2W 0.4x0.4x1.1) using Utrecht atlas with manual -N flag affine coregistration for T1 to the template (QA screenshot attached). ASHS worked perfectly on my pilot young adult data with same acquisition parameters and Magdeburg atlas.

-Why do you think my T1 MP2RAGE image quality becomes so poor within ASHS (see comparative preASHS screenshot)? It also happened to my ASHS-successful young adult data so I guess it doesn't matter as long as coreg to template still works?

-When would you recommend using NLMDenoise (or ANTs DenoiseImage if it is equivalent?) on input images for ASHS - if they have poor clarity? I always use N4 BFC, and have played around with denoising, sometimes it improves segmentations but often not. Is there another preprocessing step we could explore to boost ASHS?

-Config settings - from Wisse et al 2016 it seems the Utrecht atlas is T1W 1mm iso and T2W 0.35iso upsampled from 0.7iso. Given my own acquisition parameters above, how important is it for ASHS performance that I alter settings in ASHS_config.sh or in the atlas ASHS_user_config.sh to match the template?
e.g. the ASHS_TSE_ISO_FACTOR in ASHS_config is 100x100x500% for default 0.4x0.4x2 T2. In the ASHS_user_config for Utrecht atlas the factor is 100x100x100 (I assume because the T2 template is already iso.) So for my 0.4x0.4x1.1 T2s, should I alter the ASHS_config TSE_ISO factor to be e.g. 100x100x200 and leave the ASHS_user_config alone? Or would this not affect things too much? Are there other settings which are important to alter?

Would be really grateful for any comments to help guide my troubleshooting/sanity checks!

Best,

Coco