Classification Of Digital Angiograms Using Artificial Neural Networks

In this paper, the primary results of using neural networks for classification of digital angiograms are presented. A multilayer perceptmn was implemented and trained to identify the artery from its background. The network was trained by a back-propagation learning algorithm. The segmentation results on a digital subtraction angiogram and a cineangiogam are demonstrated. These results are also compared with the traditional maximum likelihood classifier.