AUTOMATIC PATTERN CLASSIFICATION AND STEREOSCOPIC VISION IN MEDICAL IMAGING

Medical imaging is a generic term used to define the use of medical practices to create images of the human body for clinical purpose. Today it includes a wide range of different techniques and these have greatly enhanced the quantity and quality of information available in the clinical practice. The clinician may now obtain a comprehensive view of internal structures of the human body, such as heart, kidney, lung, gut and so on. Computer assistance plays a relevant role in all these clinical applications. Each imaging technique is indeed associated with some kind of specialized workstation which maintains the appropriate tools for manipulating images, performing measurements and extracting relevant information from the available data. The major strength in the application of computers to medical imaging hence is the use of Computer Vision and Image Processing techniques to automate some specific analysis tasks. Among the thousands of possible areas, in this dissertation we exploit the current Computer Vision technologies to propose new methods in two research fields: Automatic Classification of Frames from Wireless Capsule Endoscopy and Depth Estimation in Bronchoscopic Intervention . In both cases the exploration of tubular internal structures of the human body through the analysis of endoscopic images asks for innovative and smart algorithms to translate the rough image data into useful information for the doctors.