Comparison study of clinical 3D MRI brain segmentation evaluation

Although numerous methods to segment brain MRI for extraction of white matter, gray matter and cerebrospinal fluid (CSF) have been proposed for the past two decades, little work has been done to evaluate and compare the performance of different segmentation methods on real clinical data sets, especially for CSF. This study focuses on the comparison of the four following methods for segmentation of cerebral brain MRI: gray levels thresholding, three-dimensional level set, fuzzy connectedness and FSL. Quantitative evaluation of segmentation accuracy was performed with comparison to manual segmentation on a database of 10 adult subjects.