BACKPROPAGATION NETWORK AND ITS CONFIGURATION FOR BLOOD-VESSEL DETECTION IN ANGIOGRAMS

A neural-network classifier for detecting vascular structures in angiograms was developed. The classifier consisted of a multilayer feed-forward network window in which the center pixel was classified using gray-scale information within the window. The network was trained by using the back-propagation algorithm with the momentum term. Based on this image segmentation problem, the effect of changing network configuration on the classification performance was also characterized. Factors including topology, rate parameters, training sample set, and initial weights were systematically analyzed. The training set consisted of 75 selected points from a 256 x 256 digitized cineangiogram. While different network topologies showed no significant effect on performance, both the learning process and the classification performance were sensitive to the rate parameters. The best result was obtained with a small learning rate (0.05) and a medium momentum rate (0.5). The three-layer (121-17-2) network was adequate for the problem and showed good generalization to the entire cineangiogram and other images including-direct video angiograms and digital subtraction angiograms. In a comparative study, the network demonstrated its superiority in classification performance; its classification accuracy was 92%, as compared to 68% from a maximum likelihood estimation method and 83% from a method based on iterative ternary classification. It was also shown that the trained neural-network classifier was equivalent to a generalized matched filter with a nonlinear decision tree.