Assessment of automated brain structures segmentation based on the mean-shift algorithm: Application in brain tumor

Brain tomographic techniques, such as MRI provide a plethora of pathophysiological tissue information that assists the clinician in diagnosis, therapy design/monitoring and surgery. Robust segmentation of brain tissues is a very important task in order to perform a number of computational tasks including morphological measurements of brain structures, automatic detection of asymmetries and pathologies, and simulation of brain tissue growth. In this paper we present brain structure segmentation results based on our implementation of the mean-shift algorithm and compare them with a number of well-known brain-segmentation algorithms using an atlas dataset as ground truth. The results indicate that the mean-shift algorithm outperforms the other methods. Last, the value of this algorithm in automatic detection of abnormalities in brain images is also investigated.