Novel Approach for Mass Detection in a Mammographic Computer-Aided System

X-ray images of the breast must be carefully evaluated to identify early signs of cancerous growth. Mass lesion detection is a challenging task, and in order to help radiologists in their identification, computer aided systems have been introduced. The purpose of this paper is to present a novel approach for mass detection in a mammographic computer-aided system. The proposed approach is based on the intensity specification to segment the image and to put into evidence the suspicious parts. The detection process tries to get progressively close to the suspicious region through different ranges of scales. Although different algorithms have been proposed for such task, most of them are application dependent. The suggested approach begins with a pre-processing step followed by sequence of: segmentation, features extraction and classification. This approach is particular for two reasons:  first, a new segmentation strategy based on competition scenario is suggested; secondly, detection is performed from the coarsest segmentation to the finest segmentation using a binary tree classifier. The proposed method was applied to a series of images from the Digital Database for Screening. Preliminary results are promising; a large study using more cases is currently in progress