3D Vascular Decomposition and Classification for Computer-Aided Detection

Abstract

In this study, we propose a weighted approximate convex decomposition (WACD) and classification methodology for computer-aided detection (CADe) and analysis. We start by addressing the problem of vascular decomposition as a cluster optimization problem and introduce a methodology for compact geometric decomposition. The classification of decomposed vessel sections is performed using the most relevant eigenvalues obtained through feature selection. The method was validated using presegmented sections of vasculature archived for 98 aneurysms in 112 patients. We test first for vascular decomposition and next for classification. The proposed method produced promising results with an estimated 81.5% of the vessel sections correctly decomposed. Recursive feature elimination was performed to find the most compact and informative set of features. We showed that the selected subset of eigenvalues produces minimum error and improved classifier precision. The method was also validated on a longitudinal study of four cases having internal cerebral aneurysms. Volumetric and surface area comparisons were made between expert-segmented sections and WACD classified sections containing aneurysms. Results suggest that the approach is able to classify and detect changes in aneurysm volumes and surface areas close to that segmented by an expert.