Enhancement of wavelet-based medical image retrieval through feature evaluation using an information gain measure

Abstract

Medical image retrieval covers a major application area, providing significant assistance in medical diagnosis. In this paper, several parameters for a wavelet-based retrieval system for HRCT lung images are analyzed, in order to improve the performance of the system in terms of precision and retrieval time. The information gain measure is used to evaluate extracted features to find the features that have the most discriminating power across image disease classes. A weighted similarity metric based on the evaluation is used for image retrieval. The number of levels for wavelet decomposition is studied for 2 types of wavelets, haar and daubechies-8. A filtering criteria based on the features with the highest information gain is used to enhance the retrieval time. Experiments on HRCT lung images that cover 8 disease classes show that the weighted approach achieves about 10% improvement in the average disease class precision, and 4.75% in the average total system precision, over the unweighted approach. Retrieval time is much enhanced while still maintaining high precision ratios.