Heterogeneous medical image retrieval using multi-trend structure descriptor and fuzzy SVM classifier

Abstract

This research work contributes a system for heterogeneeous medical image retrieval usiing Multi-trend structure descriptor (MTSD) and fuzzy support vector machine (FSVM) classifier. The MTSD encodes the local level structure in the form of trends for color, shape and texture information of medical images. Experimental results demonstrate that t the fusion of MTSD and FSVM significantly increases the retrieval precis iion for heterogeneeous medical image dataset. The simplest Manhattan diistance is incorporated for measuring the similarity. The feasibility of thee proposeed system is extensively experimented on benchmark da ataset and the experimental study clearly demonstrated that proposed fusion of MTSD with Fuzzy SVM gives significantly superior average retrieval precision.