Breast lesions detection in digital mammography an automated pre-diagnosis

Abstract

Breast cancer is one of the most common causes of death in the female population worldwide and one of the most prevalent cancers among other types of cancer. An early and adequate diagnosis is a key factor for an appropriate treatment, increasing the probability of survival. In order to enhance the efficiency and effectiveness of a diagnosis, an image analysis system was implemented; its purpose was to provide support for radiologists in detection of lesions from mammograms.

Image segmentation techniques were carried out to find breast lesions within the mammograms in the region of interest (ROI), which is related to the area where breast density is concentrated. Breast density is defined as the brightest part on the mammographic image and it is composed by glandular and adipose tissue where breast lesions are likely to be exposed. This study provides a methodology divided in two main segmentation techniques:

1) a region growing technique and 2) split and merge technique. This study also gives a complete description of image analysis and the tools used in