The human electrocardiogram (ECG) provides information of the heart electrical activity using electrodes placed on the skin. It is represented by a non-linear quasi-periodic time series, and it is the key indicator to examine the electrical functions and conditions of the heart. However, the fidelity of the ECG signals is often severely degraded by noise, which might alter the morphological features and the time interval aspects of the ECG leading to false diagnoses and inadequate treatment to patients. The process of denoising an ECG signal is an important pre-stage, which attenuates the noises to be able to retrieve ECG signals without changes in its morphology and time interval.
In this study, we focused on the reduction of powerline interference in ECG signals using a combined approach. We discuss the noise characteristics of typical ECG recordings and describe two filtering techniques that were adapted to process ECG measurements. We propose to use a model-based approach combined with Wiener filtering to improve ECG denoising performance. Our method was evaluated using an experimental system based on unconventional ECG electric-field-based sensing technology. Testing was performed using both real ECG signals and noise and ECG recordings from MIT-MIH arrhythmia database. The performance of our approach was evaluated using qualitative criteria (i.e., power spectrum density) and quantitative criteria (i.e., signal to noise ratio and mean square error). Finally, a comparison between our proposed methodology and existing filtering methods was carried out. We show that our approach can lead to an overall improved noise cancelling and a corresponding improved retrieval of the desired ECG signal.
Note: this abstract is for the special session "Unconventional ECG electrodes" led by Prof. Danilo Pani. Otherwise, I am happy this to be considered as regular paper