Find in Library

Background: Using long-term electrocardiogram (ECG) recorders for diagnosing cardiac diseases is only feasible by using automatic algorithms. In order for algorithms to have such reliability, they should be able to precisely detect the critical points of the ECG signal. Methods: This paper introduced a gradient-based algorithm to detect Q, R, S, P, and T points. This algorithm detects QRS-onset and QRS-offset, which are respectively the first and the last parts of QRS complex, by counting the number of threshold crossings of the slope signals. Fiducial points can then be found considering the obtained information about limitations of intervals, amplitudes and shape of the components, adaptive thresholds, and search-back algorithm. To use this algorithm in real-time telemonitoring, it is implemented on an ARM microprocessor which is fast and consumes a low amount of energy. Findings: The algorithm was implemented on MIT-BIH ECG database. The accuracy of R-point detection was 97.18% in this database. Conclusion: Since the developed algorithm is knowledge-based and sufficiently fast, it can be used in real-time software or ARM microprocessors to detect arrhythmias from ECG signals with considerably high performance.