Spherical SOM and its application to Fundus Image Analysis

Abstract

A new clustering analysis using spherical SOM has been proposed to create a more accurate representation of the data by removing the "border effect". The method is able to visualize a multi-dimensional dataset as a graphical object and to extract a tree structure from the object as a dendrogram. In my work, to better understand the potential effectiveness of the method, the clustering method using spherical SOM has been applied to real medical treatment data and has been compared the results of another approaches for clustering, which are Learning Vector Quantization, Support Vector Machine and Discriminant Analysis.