Filtering ECG Signal to Optimize the R Peak Detection Using Scipy Library with python

Abstract

In 2019, approximately 17.9 million people died because of cardiovascular disease (CVD). In the clinical immersion conducted at Lira Medika, Heart Failure (HF) is a dangerous disease that needs proper health monitoring. The common technology used to monitor heart disease is the electrocardiogram (ECG). But the concerns of electrocardiogram (ECG) raw data are hardly examined by medical experts due to noise thus troublesome to bring medication. Therefore, the research conducted to develop self-health monitoring apps UI and data flow with real-time ECG filtering features could help to monitor HF patients' health and reduce mortality. However, the noise arises from ECG measures, such as baseline wander, power line interference, miscellaneous peripherals, and muscle contraction competitively interrupted with the ECG signal. Therefore, the unfiltered ECG signal might interfere with the signal reading which leads to misinterpretation or wrong diagnosis. The Butterworth bandpass filter was suggested to provide the filtering treatment by attenuating the signal frequencies to flatten the signal. The performance of the Butterworth bandpass filter parameter was previously studied and showed a great result in attenuating the noise. In this research, a physionet database containing ECG signal type Modified Limb 2 (MLII) was used to perform a Butterworth bandpass signal filtering trial. Adjusting the high-cut and low-cut, signal frequencies could attenuate and filter the noise of the ECG signal. Therefore, the trial to test the Butterworth bandpass filter was performed and as expected provide reliable ECG data for a medical expert diagnosis with ±96,06% accuracy with ±3,94% error.