Segmentação da glândula prostática em imagens de ressonância magnética utilizando conhecimento a priori da forma e método Level Set

Abstract

Magnetic resonance imaging (MRI) is an important tool for medical diagnosis, since it is a non-invasive method and produces detailed images of physiological structures analyzed. Considering the importance of MRI for the diagnosis of several pathologies related to the prostate, it is possible to use image processing and analysis techniques in order to provide improvements in the quality of the MRI, as well as to assist in the identification and separation of the prostate gland. In this sense the objective of this work is the application of the Level Set, Free Level Set and Restrict Level Set segmentation methods in a Magnetic Resonance Imaging (MRI) repository of the male pelvic region, proposing the segmentation of the prostate gland. In order to meet the objective, a research was done in the area of image segmentation, studying the main techniques of segmentation. A repository of MRIs from the pelvis of several patients was acquired, followed by manual selection of the MRIs suitable for segmentation, validating the segmentation results using the measured Haussdorff and F-Measure distance metrics from the ground truth images. The results of the metrics confirmed the Restrict Level Set method as the most efficient, due to the better segmentation of the prostate gland in MRIs.