A reproducibility evaluation of the effects of MRI defacing on brain segmentation.

The effects of defacing are noticeable and should not be disregarded. Extra attention, in particular, should be paid to the possibility of catastrophic failures. It is crucial to adopt a robust defacing algorithm and perform a thorough quality check before releasing defaced datasets. To improve the reliability of analysis in scenarios involving defaced MRIs, it's encouraged to include multiple brain segmentation pipelines.