Topological Fragmentation of Medical 3D Surface Mesh Models for Multi-Hierarchy Anatomical Classification

Abstract

High resolution 3D mesh representations of patient anatomy with appendant functional classification are of high importance in the field of clinical education and therapy planning. Thereby, the anatomical classification can seldom be directly derived from patient morphology, thus necessitating tool support. In this work a hierarchical mesh data model for multi hierarchy anatomical classification is introduced, allowing labeling of a patient model according to various medical taxonomies. The classification regions are thereby specified utilizing a spline representation to be specified by the medical expert at low effort. Furthermore, application of randomized dilation allows conversion of the specified regions on the surface into fragmented and closed sub-meshes composing the entire anatomical structure. As proof of concept, the semi-automated classification functionality is implemented for VTK library and visualization of the multi-hierarchy anatomical model is validated for OpenGL, successfully extracting sub-meshes of the brain lobes and preparing classification regions according to Brodmann area taxonomy.