ACOUSTICAL DETECTION OF CORONARY OCCLUSIONS USING NEURAL NETWORKS

A nonlinear neural network classifier was applied to noninvasive acous tic detection of coronary artery disease; the classifier included a fe ature vector, derived from diastolic heart sounds, and a multi-layered network trained by the backpropagation. The feature vector is based o n the linear prediction coefficients of the autoregressive method afte r an adaptive line enhancement method was used as the input pattern to the neural network. One hundred and twelve recordings (70 abnormal, 4 2 normal) were studied and the network was trained on a randomly chose n set of six abnormal and six normal patients. It was tested on a data base consisting of 100 recordings to which it had not been exposed. Th e network correctly identified 50 of the 64 patients with coronary art ery disease and 32 of the 36 patients without any coronary artery occl usions. These results showed that this neural network is capable of di stinguishing normal patients from abnormal patients. In addition, the diagnostic capability of this approach is much better than any other a vailable noninvasive approach.