Features detection and navigation on neurovascular trees

摘要

The 3D reconstruction of the cerebral vascular anatomy can greatlyimprove the diagnosis of vascular pathologies, such as embolisms andhaemorrhages. This paper presents a new method for the automaticextraction of the blood vessels surface and the labelling of itsfeatures, specifically branching, aneurysms and stenoses. The methodworks with segmented magnetic resonance angiography images. It extractsa discrete medial axis transform from the stacked images and vectorizesit. During the vectorization process, the features are detected andlabelled automatically. The final output is a labelled continuous medialaxis transform representation of the vascular tree suitable forinteractive visualizations and navigations, along with 3D geometricaland topological queries