Development and benchmarking of algorithm for heartbeat detection

Abstract

This paper introduces an algorithm for ECG beat detection for portable monitoring systems using the band-pass filtering technique. It uses the Mexican-Hat wavelet as the impulse response of the corresponding band-pass filter. It relies on the adaptive threshold technique to minimize the number of false positive beats and therefore guarantee robustness against high motion artifact noise levels. The algorithm was implemented using MATLAB and has a sensitivity Se = 99.37% and a positive predictivity P+ = 99.83% with the MIT-BIH arrhythmia database. Furthermore, in order to test the performance of the algorithm against motion artifact noise, a noise stress test was performed by adding motion artifact noise to the ECG records of the same database at various signal to noise ratio (SNR) values. The results of the noise stress test were benchmarked against some existing algorithms in the literature such as Pan and Tompkins' and Romero.