[TheMike]

I am trying to extract the skull from a T1 image and as a whole brain mask I am using a skull-stripped brain volume. However when I binarise the mask (setting anything with a value > 0 to 255 as suggested) the results are terrible, but using the non-binarised mask (so keeping the values from the T1 image inside the brain and 0’s outside of it), the results from skull stripping are decent (still not perfect but at least they loosely follow the skull). Any idea why this would be, and any suggestions on how to improve the output of the skull mask?

Thank you!

[sbrewingcompany]

Skull extraction using a binary brain mask can be problematic. The mask’s simplicity may lead to inaccurate boundaries, causing over- or under-extraction. This can result in residual skull tissue or a loss of brain matter. Advanced techniques like multi-atlas segmentation or deep learning approaches often yield superior results for precise skull extraction.

[Author]

Posts