Reconstruction and topological cleaning of brain MR Images

Abstract

This paper presents a new approach for topological classification and cleaning of 3D model in medical images that respects a desired topology. First a Marching Cubes-based method is applied on segmented MR Images providing thus a closed surface mesh. A topological structure (Gmap) allowing to control the topology is then generated. Lastly a classification and correction based on specific characteristics relative to volumes, such as the number of faces or exact volume size is done. This topological structure permits to handle and manage topological, geometrical and medical information inside a healthy brain and even for a brain with a tumor or a lesion.