Late potential recognition by artificial neural networks

摘要: Ventricular late potentials (LPs) are high-frequency low-amplitude signals obtained from signal-averaged electrocardiograms (ECGs) [SAECGs]. LPs useful in identifying patients prone to ventricular tachycardia (VT), spontaneous or inducible during electrophysiology testing. A combination of self-organizing and supervised artificial neural network (ANN) models was developed identify with a positive (PEP) test for negative (NEP) using LPs. We have added morphology information vector magnitude waveform an original set three time-domain features LPs, which total QRS duration (TQRSD), signal (HFLAD), root-mean-square voltage (RMSV). Pattern recognition results ANN model this feature superior the Bayesian classification based on conventional SAECG. In order increase robustness recognition, filtered offset point is randomly shifted /spl plusmn/8 ms form fuzzy training set, simulate possible error detecting also found that nonlinear transformation through hidden layer could Euclidean distance between PEP NEP patterns.