Int. J. Patt. Recog. Art. Intell., Special Issue on MR Brain Image Analysis, 1997 1 An Integrated Approach for Locating Neuroanatomical Structure from MRI 1 Lawrence H. Staib , Amit Chakraborty and James S. Duncan lawrence.staib@yale.edu chakrab@scr.siemens.com james.duncan@yale.edu Departments of Electrical Engineering and Diagnostic Radiology, Yale University 333 Cedar Street, New Haven, CT 06520-8042 Siemens Corporate Research, 755 College Road East, Princeton, NJ 08540 Abstract The wide availability of high resolution magnetic resonance images (MRI) of the brain has fa- cilitated tremendous progress in neuroscience. Accurate automated segmentation and quantifica- tion of neuroanatomical structure from such images is crucial for the advancement of understand- ing of brain morphology, both in normal variation and in disease. Gradient-based deformable surface finding is a powerful technique for locating structure in three-dimensional images. How- ever, it often suffers from poorly defined edges and noise. This paper proposes a gradient-based deformable surface finding approach that integrates region information. This makes the resulting procedure more robust to noise and improper initialization. In addition, prior shape information may be incorporated. The algorithm uses Gauss’s Divergence theorem to find the surface of a homogeneous region-classified area in the image and integrates this with a gray level gradient- based surface finder. Experimental results on synthetic and MR brain images show a significant improvement is achieved as a consequence of the use of this extra information. Further, these improvements are achieved with little increase in computational overhead, an advantage derived from the application of Gauss’s Divergence theorem. 1 Introduction Magnetic resonance imaging (MRI) allows detailed examination of the morphology of the brain at high resolution and in vivo. Three dimensional image analysis is important in this domain in order to facilitate the quantitation necessary for better understanding of normal and abnormal structure. In most cases, the analysis requires the precise identification and quantification of structures and abnormalities in the brain in terms of volume, surface area, location and shape. The study of abnormalities and the normal variation of the shape of brain structures is important in character- izing the brain and will likely to lead to an increased understanding of the normal and abnormal morphology. Brain function can be related to morphology by examining subjects with brain disor- ders and measuring behavioral correlates to morphology in order to establish structure-function relationships. Size differences have been noted in a variety of brain disorders, including, for ex- ample, the hippocampus in posttraumatic stress disorder (PTSD) [1], the temporal lobe in learning disabilities [2] and the corpus callosum in normal twins [3]. Shape can be fully characterized by curvature in an invariant way and shape differences have also been found, for example, in the 1 This work was supported in part by the National Institute of Neurological Disorders and Stroke under Grant NS 35193.