A level set method for skull-stripping MR brain images
Haihong Zhuang, D.J. Valentino · 2005
Segmenting brain from non-brain tissue, also known as skull stripping, has become an important image processing step in analyses involving image registration or cortical flattening. We developed a new function for controlling the evolution of the zero level set, which is implicitly represented by the level set function /spl Phi/, and applied it to skull-stripping magnetic resonance (MR) images of the brain. The function uses both the intensity characteristics and known morphological features of the cortex to recognize the brain surface in 2D brain MR images. We tested the algorithm on ten sets of brain MR images and evaluated the results against manual segmentation by trained neuroscientists. The algorithm currently requires manual selection of a starting point in 2D images, but it is potentially a very accurate, stable and fast method for automated skull-stripping of brain MR volumes.