Certain improvements in support Vector machine kernel using hybrid Evolutionary algorithms for Medical image classification

Abstract

With the technological advances in digital imaging large newlinecollections of medical images are generated and stored in the medical newlinedatabases X ray CT MRI PET ultrasound images are a major source of newlineanatomical and functional information needed for diagnosis research and newlineteaching With the amount of digital medical images generated the medical newlinedatabase is huge and multi varied The need to search data collections newlineefficiently requires good image retrieval systems Content Based Image newlineRetrieval CBIR helps to retrieve images from the database similar to the newlinequery image are now widely used for medical image retrieval CBIR is newlinebasically a two step process such as Feature Extraction and Feature Matching newlineFeature Extraction is the process to extract image features to a distinguishable newlineExtend Information extracted from images such as colour texture and shape newlineare known as feature vectors This process is done on both query image and newlineimages in databases Feature Matching involves comparing the features of newlineboth the images in the database newlineIncorporating adaptable techniques to process the images of varied newlinecharacteristics and classes is the biggest issue in CBIR systems In this newlineresearch the problem of retrieving medical images from a multi varied newlinedatabase is addressed The underlying objective of this research work is to newlineinvestigate the efficacy of the various classification algorithms used in content newlinebased image retrieval and proposes a novel classification algorithm newline newline