Ultrasound image discrimination between benign and malignant adnexal masses

Abstract

This thesis deals with the discrimination between benign and malignant adnexal masses through ultrasound images. This task represents one of the most challenging problems in gynecological practice. Benign adnexal masses should be treated by minimally invasive surgery whereas patients with questionable adnexal masses should be referred for primary surgery. An accurate diagnosis is crucial in order to establish the optimal management for these patients. Therefore, it is essential for the specialist to have as many tools as possible in order to distinguish between benign and malignant adnexal masses. The diagnostic techniques that are being used involve 2D images and 3D ultrasound volumes. The basic outline of such system is the following. The rst step consists of pre-processing both the images and volumes. Then, a set of characteristics is extracted. Finally, these characteristics will be used as an input to a classi cation system. The main goal of this thesis is make this system to be part of the daily clinical practice in order to validate its viability as an aid to diagnosis