Artificial neural network analysis of common femoral artery Doppler shift signals: classification of proximal disease.

The aim of this study was to apply artificial neural networks (ANNs) to the problem of the diagnosis of aorto-iliac arterial disease on the basis of the profile of the common femoral artery (CFA) Doppler flow velocity waveform. The maximum frequency envelopes obtained from the CFA of 180 subjects were used to create sets of training and testing vectors for a back-propagation ANN. The ANN had three outputs: one representing the absence of significant aorto-iliac disease (i.e., < 50% diameter stenosis), one representing the presence of a hemodynamically significant aorto-iliac stenosis (i.e., 50-99% stenosis), and the other representing the presence of an aorto-iliac occlusion. After training, the ANN correctly classified 80% of "no significant disease" testing data, 45% of "significant stenosis" data and 85% of "occlusion" data. This work, thus, demonstrated the ability of an ANN to identify the severity of aorto-iliac disease from the CFA waveform. Although the ANN outperformed standard univariate methods and visual classification of the data, it would appear that further work is needed to increase the accuracy of the ANN to a clinically acceptable standard.