Generating Geometric Models through Self-Organizing Maps

A new algorithm for generating shape models using a Self-Organizing map (SOM) is presented. The aim of the model is to develop an approach for shape representation and classification to detect differences in the shape of anatomical structures. The Self-Organizing map requires specification of the number of clusters in advance, but does not depend upon the choice of an initial contour. This technique has the advantage of generating shape representation of each cluster and classifying given contours simultaneously. To measure the closeness between two contours, the area difference method is used. The Self-Organizing map is combined with the area difference Procrustes method and is applied to human heart cardiac borders. The experimental results show the effectiveness of the algorithm in generating shape representation and classification of given various human heart cardiac borders.