A novel and lightweight P, QRS, and T peaks detector using adaptive thresholding and template waveform. (May 2021)

• Record Type: Journal Article
• Authors: Rahul, Jagdeep; Sora, Marpe; Sharma, Lakhan Dev
• Is Part Of: Computers in biology and medicine. Volume 132(2021)
• Publication Date: 2021-05
• Subjects: Clustering -- P wave -- QRS complex -- T wave -- Template waveform; Medicine -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; 610.285
• DOI: 10.1016/j.compbiomed.2021.104307
• Languages: English
• ISSNs: 0010-4825

Abstract

Abstract: Accurate detection of key components in an electrocardiogram (ECG) plays a vital role in identifying cardiovascular diseases. In this work, we proposed a novel and lightweight P, QRS, and T peaks detector using adaptive thresholding and template waveform. In the first stage, we proposed a QRS complex detector, which utilises a novel adaptive thresholding process followed by threshold initialisation. Moreover, false positive QRS complexes were removed using the kurtosis coefficient computation. In the second stage, the ECG segment from the S wave point to the Q wave point was extracted for clustering. The template waveform was generated from the cluster members using the ensemble average method, interpolation, and resampling. Next, a novel conditional thresholding process was used to calculate the threshold values based on the template waveform morphology for P and T peaks detection. Finally, the min-max functions were used to detect the P and T peaks. The proposed technique was applied to the MIT-BIH arrhythmia database (MIT-AD) and the QT database for QRS detection and validation. Sensitivity (Se%) values of 99.81 and 99.90 and positive predictivity (+P%) values of 99.85 and 99.94 were obtained for the MIT-AD and QT database for QRS complex detection, respectively. Further, we found that Se% = 96.50 and +P% = 96.08 for the P peak detection, Se% = 100 and +P% = 100 for the R peak detection, and Se% = 99.54 and +P% = 99.68 for the T peak detection when using the manually annotated QT database. The proposed technique exhibits low computational complexity and can be implemented on low-cost hardware, since it is based on simple decision rules rather than a heuristic approach. Highlights: The proposed method is a computationally lightweight P, QRS, and T peaks detector. Our method utilises adaptive thresholding for QRS complex detection and template waveforms for P and T peaks detection. The proposed method has successfully achieved Se% = 99.81 and +P% = 99.85 for QRS complexes detection on MIT-AD. Our approach was also validated on the manually annotated QT database for P, R, and T peaks detection. The proposed method shows significantly better performance compared with other existing state-of-the- art algorithms.