I'm using the spm12 version of what used to be called "new segment" with a non standard template (neonate template, 9 tissue classes) on T2 images. Everything functions as expected with spm 8, however there is some strange behaviour with spm12 that I'm trying to track down. The issue is a region of misclassification, in csf near the bottom of the brain stem. That area gets misclassified as tissue class 5, irrespective of what tissue class 5 is. In the original template class 5 is the hippocampus, but if I switch classes 1 (cortical GM) and 5, then the misclassification is as cortical GM. There is nothing unusual about the patch of csf that gets misclassified that I can find. It is not unusually bright or unusually dark. It is certainly more similar to the rest of the csf than the class it gets misclassified as.

I recall reading somewhere that there is something special to do with the eyes in spm12, but I haven't been able to track down differences in code between spm8 and spm12 that might account for the difference in behaviour. Is there some special hidden option of hard coding that makes class 5 behave differently?

Any advice appreciated.

Thanks

Richard Beare

Developmental Imaging

Murdoch Childrens Research Institute

Melbourne

Australia.