Research on indexing in medical image database retrieval using self-organizing maps

Abstract

High dimensional indexing scheme is a significant research issue for content-based image retrieval in medical image database. The SOM-based approach proposed in this paper uses a Kernel Density Estimation Method (Mean Shift) to describe for medical image database, which fits the complex data distribution reasonably well. And this approach trains optimized SOM network to partition data space. Experiments on a real-world medical image dataset demonstrate a good topology configuration of the data and a remarkable reduction of the amount of accessed vectors in exact NN searches compare with existing indexing schemes.