Detection of QRS-duration in Chagas' patients with wavelets

Individuals with Chagas' disease develop abnormal ventricular conduction patterns which can be measured with late potentials (LP) measured from signal-averaged ECG (SAECG). The end-point of the QRS (and thus total QRS duration-TQRSD) must be accurately detected in order to determine if an individual exhibits LP. The authors have found that the standard time-domain method under estimates TQRSD in noisy signals. Their hypothesis is that the TQRSD can be more accurately determined from the wavelet-transformed SAECG in a neural network in noisy data. The authors compared the TQRSD measured by experts with the standard method and the authors' proposed method. The results of the proposed method were highly correlated with those from the experts (Slope=0.94, Int.=7.1 ms, R=0.94). The proposed method reduced the difference between the automatically, determined and manually determined values by 57%. It is concluded that the proposed method has good potential to help automate the analysis of cardiac conduction anomalies in chagasic patients.