A neural network approach for patient-specific 12-lead ECG synthesis in patient monitoring environments

In recent years, there has been a growing interest in developing accurate methods for the synthesis of the 12-lead ECG from a minimal lead-set to improve patient monitoring in situations where the acquisition of the 12-lead ECG is difficult or impractical. This paper presents a method that aims to derive the standard 12-lead ECG from a pseudoorthogonal 3-lead subset via a nonlinear patient specific reconstruction method that is based on the use of artificial neural networks (ANN). We train and test the ANN over a 300 adult patients study population. We then assess the performance of the ANN based ECG synthesis method in comparison with the multiple regression based method and test for statistical differences between the two methods using the paired Student's t-test. The ANNs achieved high overall accuracies for all the testing sets. Moreover, the difference in accuracies between both methods is statistically significant (p<0.001). The encouraging results reported here suggest that the artificial neural networks represent a rather interesting and very promising approach to improve the synthesis of the 12-lead ECG.