Best segmentation routine for CSF Volume count in children with hydrocephalus

Dear all,
I just discovered this platform. I am currently helping a neuropediatric radiologist to develop a pipeline for CSF Volume quantification before and after shunt surgery. We have tried spm and cat12 but in roughly 50% of the cases these algorithms fail, because of the disproportionate intracranial anatomy of the children that don’t match any reasonable tpm. I am thinking about ways for only signal based segmentation in native space. Does anyone know of such algorithms or software solutions?

Any help is mich appreciated.

Best wishes