Development of tolerant features for characterization of masses in mammograms

In this paper, four new features for the analysis of breast masses are presented. These features were designed to be insensitive to the exact shape of the contour of the masses, so that an approximate contour, such as the one extracted via an automated segmentation algorithm, can be employed in their computation. Two of the features, Sp(SI) and Sp(GO), measure the degree of spiculation of a mass and its likelihood of being spiculated. One of these features, Sp(GO), is a measure of the relative gradient orientation of pixels that correspond to possible spicules. The other feature, Sp(SI), is based on a comparison of mutual information measures between selected components of the mammographic images. The last two features, Fz(1) and Fz(2), measure the local fuzziness of the mass margins based on points defined automatically. The features were tested for characterization (i.e. discrimination between circumscribed and spiculated masses) and diagnosis (i.e. discrimination between benign and malignant masses) of breast masses using a set of 319 masses and three different classifiers. In the characterization experiments the features produced a result of approximately 89% correct classification. In the diagnosis experiments, the performance achieved was approximately 81% of correct classification.