Automatic p wave analysis over 24 hours in patients with paroxysmal or persistent atrial fibrillation.

BACKGROUND: Patients with atrial fibrillation (AF) often exhibit abnormalities of P wave morphology during sinus rhythm. We examined a novel method for automatic P wave analysis in the 24-hour-Holter-ECG of 60 patients with paroxysmal or persistent AF and 12 healthy subjects. METHODS: Recorded ECG signals were transferred to the analysis program where 5-10 P and R waves were manually marked. A wavelet transform performed a time-frequency decomposition to train neural networks. Afterwards, the detected P waves were described using a Gauss function optimized to fit the individual morphology and providing amplitude and duration at half P wave height. RESULTS: >96% of P waves were detected, 47.4 +/- 20.7% successfully analyzed afterwards. In the patient population, the mean amplitude was 0.073 +/- 0.028 mV (mean variance 0.020 +/- 0.008 mV(2)), the mean duration at half height 23.5 +/- 2.7 ms (mean variance 4.2 +/- 1.6 ms(2)). In the control group, the mean amplitude (0.105 +/- 0.020 ms) was significantly higher (P < 0.0005), the mean variance of duration at half height (2.9 +/- 0.6 ms(2)) significantly lower (P < 0.0085). CONCLUSIONS: This method shows promise for identification of triggering factors of AF.

Measuring nonstationarity by analyzing the loss of recurrence in dynamical systems.

We propose a measure for nonstationarity which is based on the analysis of distributions of temporal distances of neighboring vectors in state space. As an extension of previous techniques our method does not require a partitioning of the time series. Moreover, the deviation of mean recurrence times from frequency distributions that would be expected under stationary conditions allows us to estimate the statistical significance of the method.

Automatic pattern recognition in ECG time series.

In this paper, a technique for the automatic detection of any recurrent pattern in ECG time series is introduced. The wavelet transform is used to obtain a multiresolution representation of some example patterns for signal structure extraction. Neural Networks are trained with the wavelet transformed templates providing an efficient detector even for temporally varying patterns within the complete time series. The method is also robust against offsets and stable for signal to noise ratios larger than one. Its reliability was tested on 60 Holter ECG recordings of patients at the Department of Cardiology (University of Bonn). Due to the convincing results and its fast implementation the method can easily be used in clinical medicine. In particular, it solves the problem of automatic P wave detection in Holter ECG recordings.