[QUESTION] Preprocessing pediatric brain data

Hello,

My group would like to evaluate your model on our internal dataset. However, when performing registration from the MRI sequences onto the SRI24 atlas, following the procedure described in your paper and using the brain-lesion-preprocessing repository, we are encountering noticeably choppy results. This may be expected, as pediatric brain volumes are smaller than the adult SRI24 atlas, and the default in the preprocessing pipeline appears to use rigid registration for atlas alignment. We've thought of using affine registration to allow the brain to scale, however, we don't want to distort the brain.

Could you please clarify the exact registration procedure or parameters you used for this step? If possible, would you be open to sharing the script or configuration you used?

Thank you for your time.