Automatic brain mr image segmentation by relative thresholding and morphological image analysis

We present an automatic method for segmentation of white matter, gray matter and cerebrospinal fluid in T1weighted brain MR images. We model images in terms of spatial relationships between near voxels. Brain tissue segmentation is first performed with relative thresholding, a new segmentation mechanism which compares two voxel intensities against a relative threshold. Relative thresholding considers structural, geometrical and radiological a priori knowledge expressed in first-order logic. It makes intensity inhomogeneity transparent, avoids using any form of regularization, and enables global searching for optimal solutions. We augment relative thresholding mainly with a series of morphological operations that exploit a priori knowledge about the shape and geometry of brain structures. Combination of relative thresholding and morphological operations dispenses with the prior skull stripping step. Parameters involved in the segmentation are selected based on a priori knowledge and robust to inter-data variations.