Global and Local Shape Analysis of the Hippocampus Based on Level-of-Detail Representations

Abstract

Both volume and shape of the organs within the brain such as hippocampus indicate their abnormal neurological states such as epilepsy, schizophrenia, and Alzheimer's diseases. This paper proposes a new method for the analysis of hippocampal shape using an integrated Octree-based representation, consisting of meshes, voxels, and skeletons. Initially, we create multi-level meshes by applying the Marching Cube algorithm to the hippocampal region segmented from MR images. Then, we convert the polygonal model to intermediate binary voxel representation by a depth-buffer based voxelization, which makes it easier to extract a 3-D skeleton as well as relate to original MR images. As a similarity measure between the shapes, we compute L2 norm and Hausdorff distance for each sampled mesh by shooting the rays fired from the extracted skeleton. It also allows an interactive analysis because of the octreebased data structure. Moreover, it increases the speed of analysis without degrading accuracy by using a hierarchical level-of-detail approach. © Springer-Verlag 2004.