ECG Denoising Using Multiple Approaches

Today, heart diseases and cardiac-affiliated failures are one of the main causes of people losing their lives in the world. Cardiovascular diseases (CVDs) are a classification of diseases that deals with the heart and blood vessels. Electrocardiogram (ECG/EKG) is an efficient and a non-intrusive index for the classification, prevention, and detection of arrhythmia. These signals are contaminated by various noises like Muscle Artifact, Baseline Wander, Power Line Interference (PLI), etc., which may lead to wrong interpretations about the patient's condition and will affect the diagnosis and monitoring. Therefore, ECG denoising is one of the major pre-processing steps for the attenuation of these noises. In the proposed work, five different techniques namely a Denoising Autoencoder (DAE) Discrete Wavelet Transform (DWT), Butterworth Filter, FIR Filter, and Median Filter. The denoising performance is estimated using RMSE and SNR. The above-mentioned techniques are compared, analyzed and results show that the proposed Denoising Autoencoder performs better in terms of SNR and RMSE and is more efficient.