Hierarchical neural network system to breasts cancer diagnosis

This paper describes a hierarchical neural network system used as a help tool in the breast cancer diagnosis. The system divides the image using a tree scheme. The preprocessing of the areas pointed by the tree takes into account the environment that the element to analyze have, so as others computations, that are done using an adaptive filter. The neural network is in charge of classifying the areas into two possible classes, a healthy class or an ill class, using the obtained descriptors from the areas. All the data employed in this research have been obtained from the digitalization of the mammographies taken to some hospitalized patients.