Development of analytical approach for an automated analysis of continuous long-term single lead ECG for diagnosis of paroxysmal atrioventricular block

Reliable detection of significant ECG features such as the P-wave, QRS-complex and T-wave are of major clinical importance. In this paper we introduce a new algorithm based on synchrosqueezing wavelet transform for detection of P-waves in long-term ECG recordings. Synchrosqueezing is a powerful time-frequency analysis tool that provides precise frequency representation of a multicomponent signal through mode decomposition. First, we analyzed four wavelet filters with different filter parameters, to identify the best specification for quantification of QRS and P-wave. Second, the algorithm was tested on ECG recording comprising of events with paroxysmal atrioventricular block and validated through visual scanning. Using morlet wavelet with a peak frequency of 5Hz and separation of 0.1Hz, our proposed algorithm was able to detect 95.5% of P-waves. From this study, it appears that synchrosqueezing wavelet transform may provide a powerful robust technique for automated ECG analysis.