Online robust R-peaks detection in noisy electrocardiograms using a novel iterative smart processing algorithm

Abstract Nowadays, many contributions deal with R-peak detection in Electrocardiographic (ECG) signals. Although they present an accurate performance in detection, most of these are presented as offline solutions, both to be processed in high performance platforms (under a big cost), or to be analyzed in laboratories without constraints in time, neither in computational load. Owing to this, it is also very important to take one step further, trying to develop new solutions which work in portable/wearable low-cost platforms, with constraints in time and in computational load. In this work, an accurate and computationally efficient method for online and robust detection of R-Peaks is presented. This method is divided in three main stages: first, in the pre-processing stage, a complete elimination of artifacts is performed based on a noise and signal intensity approach; second, R-peaks detection is carried out through an efficient “area over the curve” method; finally, in the third stage, a novel iterative algorithm consisting in three sequential state machines performs the correct detection of the R-peaks applying heart period distance rules. Moreover, the method is performed over time in short length sliding windows. The algorithm has been tested using all 48 full-length ECG records of the MIT-BIH Arrhythmia Database, achieving 99.54% sensitivity and 99.60% positive predictivity in R-peak detection.