Unsupervised Skull Stripping in MRI

Whole brain segmentation, referred to as skull stripping, is an important technique in neuroimaging. Many applications, such as presurgical planning, cortical surface reconstruction and brain morphometry, depend on the ability to accurately segment brain from non-brain tissue, i.e. remove extra-cerebral tissue such as skull, sclera, orbital fat, skin, etc. However, despite the clear definition of this essential step, no universal solution has been developed that is robust to neuroanatomical variability and the types of noise present in standard structural MRI sequences. We approach the skull-stripping problem by combining watershed algorithms and deformable surface models. Our method takes advantage of the simplicity and robustness of the former, while using the accuracy and surface information available to the latter. A training set of accurately segmented brains validates the segmentation and eventually corrects it, resulting in a robust and automated procedure. Thesis Supervisor: Olivier Faugeras Tide: Adjunct Professor of E.E.& C.S. Thesis Supervisor: Bruce Fischl Title: Assistant Professor of Radiology