Improving GRNNs in CAD systems

Abstract

Different Computer Aided Diagnosis (CAD) systems have been recently developed to detect microcalcifications (MCs) in digitalized mammography, among other techniques, applying General Regression Neural Networks (GRNNs), or Blind Signal Separation techniques. The main problem of GRNNs to achieve an optimal classification performance, is fitting the kernel parameters (KPs). In this paper we present two novel algorithms to fit the KPs, that have been successfully applied in our CAD system achieving an improvement in the classification rates. Important remarks about the application of Gradient Algorithms (GRDAs) are assessed. We make a brief introduction to our CAD system comparing it to other architectures designed to detect MCs. © Springer-Verlag 2004.