Discrimination between benign & malignant breast tumors using a region-based measure of edge profile acutance

Abstract

We propose an adaptive measure of edge profile acutance to distinguish between benign and malignant breast tumors on mammographic images. The measure is computed using differences along normals to the boundary of the tumor region. Classification rates of up to about 95% were obtained in two-group and four-group classification (the latter using additional shape factors) of 39 tumors from the MIAS database and 54 tumors in an augmented database.