Supporting Content Based Visual Information Retrieval for Medical Imaging With Lenient Relevance Feedback

Abstract

ABSTRACT newlineDue to the advancement of digital imaging technology, the concept of medical image processing flourished. The medical image processing deals with several kinds of medical images and these images help in easy diagnosis of the abnormality. The medical images are exploited to visualize the internal body, to track and to diagnose the presence of any abnormality being present. Medical images make the process of diagnosis without any hassles and so, it is actively employed in the medical world. newlineThe employment of medical images demand better processing of images to arrive at reliable decisions. There are different medical imaging modalities and is chosen by considering the part of the body and the physical condition of the patient. Interpretation of medical images require advanced image processing techniques. Reliability and accuracy are the two significant requirements of any medical image processing system. As medical image processing applications are closely linked to the human lives, inaccuracy leads to major issues. newlineThe usual applications of medical image processing include medical image segmentation, noise removal, medical image classification and so on. Content Based Image Retrieval (CBIR) is one of the popular applications of digital image processing, but it is found scarce in the applications related to medical images. The objective of CBIR is to retrieve relevant medical images from the medical database with reference to the query image in a shorter span of time. newlineThe medical CBIR systems take a medical image as input and search the entire medical database to detect the medical images of similar kind. This activity helps in analysing the relevant images with respect to the query image and the diagnostic procedures can be reviewed easily. Hence, it is easy for the healthcare professional to review and analyse the similar cases. newlineThough there are numerous medical image processing applications, the applications meant for kidney are observed to be limited. Hence, this work attempts to present three different CBIR systems for ultrasound images of kidney by incorporating image processing and data mining concepts. All the proposed approaches are different, yet the research goal is to attain better accuracy in a reasonable amount of time. newlineThe initial phase of this research presents a feature selection technique that aims to improvise the medical image diagnosis by selecting prominent features. As soon as the features are selected, association rules are formed and the kidney images are classified. The second phase of the research extracts features and the association rules are formed by the proposed Classification Based on Highly Strong Association Rules (CHiSAR). Finally, the rule subset classifier is employed to classify between the images. newlineThe final phase of the research extracts the features from the kidney images and the association rules are reduced for better performance. The image relevance inference is performed and finally, binary and the best first search classification is employed to classify between the images. The performances of all the proposed approaches are analysed in terms of accuracy and time consumption. The proposed approaches prove better results with better accuracy and reasonable time consumption. newline newline