Applying Pulse Spectrum Analysis to Facilitate the Diagnosis of Coronary Artery Disease

Abstract

Not all patients with angina pectoris have coronary artery stenosis. To facilitate the diagnosis of coronary artery disease (CAD), we sought to identify predictive factors of pulse spectrum analysis, which was developed by Wang and is one technique of modern pulse diagnosis. The patients suffered from chest pain and received cardiac catheterization to confirm the CAD diagnosis and Gensini score were recruited. Their pulse waves of radial artery were recorded. Then, by performing a fast Fourier transform, 10 amplitude values of frequency spectrum harmonics were obtained. Each harmonic amplitude was divided by the sum of all harmonic amplitude values, obtaining the relative percentages of 10 harmonics (C1-C10). Subsequently, multivariate logistic regression was conducted with two models and the areas under the receiver operating characteristic curves (ROC) of these 2 models were compared to see if combining the pulse diagnosis parameters with the risk factor of CAD can increase the prediction rate of CAD diagnosis. The predictive factors of CAD severity were analyzed by multivariate linear regression. A total of 83 participants were included; 63 were diagnosed CAD and 20 without CAD. In the CAD group, C1 was greater and C5 was lower than those of the non-CAD group. The CAD risk factors were put alone in Model 1 to perform the multivariate logistic regression analysis which had a prediction rate of 77.1%; while putting the C1 and C5 harmonics together with the risk factors into Model 2, the prediction rate increased to 80.7%. Finally, the area under ROC of Model 1 and Model 2 was 0.788 and 0.856, respectively. Furthermore, left C1, left C5, gender, and presence of hyperlipidemia were predictors of CAD severity. Therefore, pulse spectrum analysis may be a tool to facilitate CAD diagnosis before receiving cardiac catheterization. The harmonics C1 and C5 were favorable predictive indicators.