Shape matching in multimodal medical images using point landmarks with Hopfield net

Abstract

Analysis of concavities within complex patterns provides a promising approach to structural shape descriptions. Such concavities are characterized by entrance, exit and height points, collectively termed as point landmarks, as well as height and width. This paper presents a scheme for using these characteristics with the Hopfield neural network for matching structural shape descriptions corresponding to identical regions of interest in images of the same object obtained from different sensors. Application to biomedical imaging is also discussed. In these applications, both the shape description and the Hopfield network based matching scheme are illustrated for matching neuroanatomical shapes of a human brain obtained from axial slices obtained from CT and MR modalities of the same region.