Detecção de regiões de massas em mamografias usando índices de diversidade, geoestatísticas e geometria côncava

Abstract

Breast cancer is configured as a global health problem that affects mainly the female population. It is known that early detection increases the chances of an effective treatment and improves the prognosis of the disease. With this goal, computacional tools have been proposed in order to assist the physician in the interpretation of mammography features providing detection and diagnosis of lesions. The challenge is to detect any lesios with high sensitivity rate while maintaining a small number of false positives. The main objective of this research is the development of an efficient methodology for mass detection in digitized mammograms. The detection task involves aspects of computer vision like find suspicious areas and describe them in a discriminatory way. This research evaluates the approaches of feature extraction using diversity analysis, geostatistics and concave geometry for the classification of previously identified suspicious regions using Support Vector Machine as a classifier techinique. The results are promising and reaches a high sensitivity rate jointly with a low mean rate of false positives per image when using concave geometry as features extraction approach.