A Hybrid Approach Based Medical Image Retrieval System Using Feature Optimized Classification Similarity Framework

Abstract

For the past few years, massive upgradation is obtained in the pasture of Content Based Medical Image Retrieval (CBMIR) for effective utilization of medical images based on visual feature analysis for the purpose of diagnosis and educational research. The existing medical image retrieval systems are still not optimal to solve the feature dimensionality reduction problem which increases the computational complexity and decreases the speed of a retrieval process. The proposed CBMIR is used a hybrid approach based on Feature Extraction, Optimization of Feature Vectors, Classification of Features and Similarity Measurements. This type of CBMIR is called Feature Optimized Classification Similarity (FOCS) framework. The selected features are Textures using Gray level Co-occurrence Matrix Features (GLCM) and Tamura Features (TF) in which extracted features are formed as feature vector database. The Fuzzy based Particle Swarm Optimization (FPSO) technique is used to reduce the feature vector dimensionality and classification is performed using Fuzzy based Relevance Vector Machine (FRVM) to form groups of relevant image features that provide a natural way to classify dimensionally reduced feature vectors of images. The Euclidean Distance (ED) is used as similarity measurement to measure the significance between the query image and the target images. This FOCS approach can get the query from the user and has retrieved the needed images from the databases. The retrieval algorithm performances are estimated in terms of precision and recall. This FOCS framework comprises several benefits when compared to existing CBMIR. GLCM and TF are used to extract texture features and form a feature vector database. Fuzzy-PSO is used to reduce the feature vector dimensionality issues while selecting the important features in the feature vector database in which computational complexity is decreased. Fuzzy based RVM is used for feature classification in which it increases the response rate and speed of the retrieval process. This proposed FOCS framework is used to help the physician to obtain more confidence in their decisions for diagnosis and medical research students are zeal to get the essential images successfully for further investigation of their research.