A Comprehensive Design and Simulation of Denoising ECG Signals Using Adaptive Filtering and Non Local Means Patch Based Method

Abstract

ECG can be used to diagnose the stable and unstable heart disorder. If ECG is properly evaluated for its noise-free waveform, it can provide very useful heart information. Since ECG is not stationary but has signals that vary time, the impurities are not normal and thus not continuously visible. The precise findings are not always guaranteed due to the existence of artifacts / noise in the waveform. In addition, visual inspection is not so accurate. The presence of artifacts / noise in waveforms often contributes to the loss of very significant and unwanted information.

Therefore it is very important to eliminate these artifacts / noise from wavelengths and to denoise the sound on the device prior to any decision. The key issue with ECG signals noise / artifact reduction is information loss when the noise is removed. The main aim of this work is therefore to define the reasonably simple way to denote ECG signals without jeopardizing the loss of information. This goal motivated me to study and experiment with the different techniques of demotion for ECG signals.

This study starts with the core operation of the heart, simple electrocardiographic presentation, multiple noise / artifacts, previous work carried out on ECG noise reduction and various noise removal methodologies. The data from the MIT-BIH database for the proposed work is also generated in MATLAB. Further information is given on the different adaptive filtering algorithms and the patch process.