Robust R-Peak Detection in Low-Quality Holter ECGs Using 1D Convolutional Neural Network

Abstract

Noise and low quality of ECG signals acquired from Holter or wearable devices deteriorate the accuracy and robustness of R-peak detection algorithms. This paper presents a generic and robust system for R-peak detection in Holter ECG signals. While many proposed algorithms have successfully addressed the problem of ECG Rpeak detection, there is still a notable gap in the performance of these detectors on such low-quality ECG records. Methods: In this study, a novel implementation of the 1D Convolutional Neural Network (CNN) is used integrated with a verification model to reduce the number of false alarms. This CNN architecture consists of an encoder block and a corresponding decoder block followed by a sample-wise classification layer to construct the 1D segmentation map of R-peaks from the input ECG signal. Once the proposed model has been trained, it can solely be used to detect R-peaks possibly in a single channel ECG data stream quickly and accurately, or alternatively, such a solution can be conveniently employed for real-time monitoring on a lightweight portable device. Results: The model is tested on two open-access ECG databases: The China Physiological Signal Challenge (2020) database (CPSC-DB) with more than one million beats, and the commonly used MIT-BIH Arrhythmia Database (MIT-DB).Experimental results demonstrate that the proposed systematic approach achieves 99.30% F1-score, 99.69% recall, and 98.91% precision in CPSC-DB, which is the best R-peak detection performance ever achieved. Results also demonstrate similar or better performance than most competing algorithms on MIT-DB with 99.83% F1-score, 99.85% recall, and 99.82% precision. Significance: Compared to all competing methods, the proposed approach can reduce the false-positives and falsenegatives in Holter ECG signals by more than 54% and 82%, respectively. Conclusion: Finally, the simple and invariant nature of the parameters leads to a highly generic system and therefore applicable to any ECG dataset.

PN-QRS: An Uncertainty-aware QRS-complex Detection Method for Wearable ECGs

“…With the aid of the proposed uncertainty quantification method, the QRS-complex location performances of PN-QRS have been close to the limit on the dynamic multi-lead ECG databases. For MIT-BIH, the F 1 score achieves 99.95%, which has exceeded the limits of the current R-peak location methods (99.86% in [43], 99.83% in [44] and 99.91% in [5]) in this database. Similarly, the cumulative errors, FPs and FNs, are only 15 and 9 for PN-QRS evaluated in INCART, approximating the limit of the QRS-complex location performance in this database.…”

Global ECG Classification by Self-Operational Neural Networks With Feature Injection

“…We found that arrhythmia beats (S and V beats) are far less frequent than the normal beats in the MIT-BIH dataset. As described in [5], we generated augmented arrhythmic rhythms from the 20-second ECG segments containing one or more arrhythmia beats by adding baseline wander and motion artifacts from the Noise Stress Test Database [55].…”

“…Additionally, we intend to expand our research on arrhythmia classification and its generalization to Holter ECGs with low-quality ECG records. In a recent study [64] we showed that the performance of R peak detection drastically decreases when algorithms that are developed for clean ECG signals are applied to noisy and lowquality Holter ECG data. For this purpose, we are planning to use the China Physiological Signal Challenge (2020) database (CPSC-DB) [65], [66], the largest Holter ECG database which contains more than one million beats.…”

“…ECG-based arrhythmia classification is typically initiated with a peak detection/segmentation. This study does not discuss R-peak detection since highly accurate algorithms have already been proposed in the literature [4], [5]. The analysis and classification of ECG signals have been extensively studied throughout the last decades [6]- [9].…”

Electrocardiograph signal recognition using wavelet transform based on optimized neural network

“…The decreased vector is then fed into an optimized neural network as a train input. The previous studies Zahid et al [15] [16] in 2019, the goal of this research is to create a real-time and accurate QRS complex detector using a new method based on the Bayesian framework. To detect QRS complexes, we present a new technique consisting of two stages: variance-based detection (VBD) and maximum-likelihood estimation (MLE).…”