Extraction of QRS fiducial points from the ECG using adaptive mathematical morphology

QRS complex detection in the electrocardiogram (ECG) has been extensively investigated over the last two decades. Still, some issues remain pending due to the diversity of QRS complex shapes and various perturbations, notably baseline drift. This is especially true for ECG signals acquired using wearable devices. Our study aims at extracting QRS complexes and their fiducial points using Mathematical Morphology (MM) with an adaptive structuring element, on a beat-to-beat basis. The structuring element is updated based on the characteristics of the previously detected QRS complexes for a more robust and precise detection. The MIT-BIH arrhythmia and Physionet QT databases were respectively used for assessing the detection performance of R-waves and other fiducial points. Furthermore, the proposed method was evaluated on a wearable-device dataset of ECGs during vigorous exercises. Results show comparable or better performance than the state-of-the-art with a 99.87% sensitivity and 0.22% detection error rate for the MIT-BIH arrhythmia database. Efficient extraction of QRS fiducial points was achieved against the Physionet QT database. On the wearable-device dataset, an improvement of more than 10% in QRS complex detection rate compared to classic approaches was obtained.