Computer Aided Diagnosis Of Gastrointestinal Disorders

Abstract

Wireless capsule endoscopy (WCE) is a standard and modern modality which newlineenables physicians to perform screening of the entire gastrointestinal tract (GIT). It is a non-invasive technique and carried out with a miniature, optical endoscope packed within a small capsule that is swallowed by the patient. The capsule endoscope advances through the GIT and simultaneously captures thousands of color images. A recorder is placed in a belt which is wrapped around patient s waist. Recorder receives images form the capsule wirelessly. The images saved on the recorder are transferred to a computer with special software that strings the images together to create a video. newlineThe challenge in this procedure lies in the last phase of the examination when newlineit needs manual evaluation of complete WCE video by an expert with keen attention. newlineDepending on the frame rate of the capsule, the images captured by the capsule ranges between 50000 to 100000. Examining such huge amount of frames with diverse newlinecontent is a complicated task. Moreover, image interpretation by human expert is newlinesubjective. Fatigue and complexity of the images, may lead to extensive variations newlineacross different interpreters. newlineThis thesis presents an efficient computer aided diagnosis systems with newlineinnovative approaches and reduced time complexity for diagnosis of gastrointestinal disorders. WCE video summarization aids efficient visualization of prolonged WCE video. Segmentation of lesions gives idea about the severity of the disease. newlineClassification of anomalies reduces the analysis time to a greater extent. The research work in the WCE analysis field is carried out in four phases: a) Collection of database; b) Summarization of WCE video; c) Pre-processing and segmentation of abnormalities; d) Classification of images into three categories such as ulcer, polyp and inflammatory. newline