QRS Detection Based on Medical Knowledge and Cascades of Moving Average Filters

Heartbeat detection is the first step in automatic analysis of the electrocardiogram (ECG). For mobile and wearable devices, the detection process should be both accurate and computationally efficient. In this paper, we present a QRS detection algorithm based on moving average filters, which affords a simple yet robust signal processing technique. The decision logic considers the rhythmic and morphological features of the QRS complex. QRS enhancing is performed with channel-specific moving average cascades selected from a pool of derivative systems we designed. We measured the effectiveness of our algorithm on well-known benchmark databases, reporting F1 scores, sensitivity on abnormal beats and processing time. We also evaluated the performances of other available detectors for a direct comparison with the same criteria. The algorithm we propose achieved satisfying performances on par with or higher than the other QRS detectors. Despite the performances we report are not the highest that have been published so far, our approach to QRS detection enhances computational efficiency while maintaining high accuracy.