No abstract

Wearable electrocardiogram (ECG) monitoring devices have enabled everyday ECG collection in our daily lives. However, the condition of ECG signal acquisition using wearable devices varies and wearable ECG signals could be interfered with by severe noises, resulting in great challenges of computer-aided automated ECG analysis, especially for single-lead ECG signals without spare channels as references. There remains room for improvement of the beat-level single-lead ECG diagnosis regarding accuracy and efficiency. In this paper, we propose new morphological features of heartbeats for an extreme gradient boosting-based beat-level ECG analysis method to carry out the five-class heartbeat classification according to the Association for the Advancement of Medical Instrumentation standard. The MIT-BIH Arrhythmia Database (MITDB) and a self-collected wearable single-lead ECG dataset are used for performance evaluation in the static and wearable ECG monitoring conditions, respectively. The results show that our method outperforms other state-of-the-art models with an accuracy of 99.14% on the MITDB and maintains robustness with an accuracy of 98.68% in the wearable single-lead ECG analysis.

Recognizing unreadable electrocardiogram (ECG) signals could reduce the error rate of automatic software analysis and improve the interpretation efficiency of doctors, especially for single-lead dynamic ECGs. In this paper, we propose an unreadable ECG segment recognition method based on morphological algorithm and random forest classifier (RFC). The single-lead ECG signals are first filtered and normalized for morphological opening and closing operation, to generate detection sequences with more obvious QRS waves, since the large amplitudes introduced by motion interference could be suppressed during this procedure. Then features such as Shannon entropy and kurtosis are extracted and the RFC is used for unreadable segment classification. A total of 3354 readable segments and 2199 unreadable segments with a length of 4 seconds are obtained from 37 patients for method evaluation. The accuracy of our method (92.94 ± 0.93%) is significantly higher than that of the method without morphological algorithm (85.68 ± 1.30%). Moreover, we also used the "N" and "~" categories of the database from PhysioNet/CinC Challenge 2017 for further verification, and the accuracy of the proposed method (93.75 ± 0.69%) is significantly higher than that of the model without morphological processing (82.25 ± 1.06%) as well.

Accurate P wave detection is important for arrhythmia diagnosis, e.g. P wave absence or P duration for atrial fibrillation diagnosis and other atrial arrhythmias. Phasor transform is an effective method for ECG fiducial points delineation. It maps each ECG sample into a phasor to enhance slight variations and preserves morphology and magnitude characteristics. In this paper, we optimized the automatic P wave delineation method based on phasor transform in four aspects, i.e., signal denoising, wave localization, candidate points detection, and optimal points selection. In our experiments, the length of the search window and the degree of phasor transform were established through various trials. Especially, along with zero-crossing points of the phasor signal, intersections of the phasor signal and the original ECG signal are obtained as candidates, which make the most contribution to delineation results. For validation, the QT Database with 3194 P wave annotations from 105 records of two leads is adopted. As a result, we reached F1 scores of 94.67% and 93.56% with detection error rates (DERs) of 10.80% and 13.06% for P wave onset and offset points detection, respectively. The F1 score and DER for P peak detection under a tolerance of 75 ms were 95.33% and 9.46%, respectively, which outperforms other reproducible works and their combinations.

The quality of ECG signals is commonly affected by severe noise, especially for the single-lead ECG signals acquired from long-term wearable devices. Recognizing and ignoring these interfered signals can reduce the error rate of automatic ECG analysis system, and in addition, improve the efficiency of cardiologists. Based on XGBoost classifier, we propose an unreadable ECG segment recognition method using features extracted through Shannon Energy Envelope (SEE) and Empirical Mode Decomposition (EMD). An unreadable CarePatchTM ECG patch database is established, containing 8169 readable segments and 6114 unreadable segments with a length of 10 seconds. The XGBoost with 5-fold cross-validation is applied and obtained an accuracy of 99.51+/-0.15%. In conclusion, SSE and EMD features contribute to the unreadable segments recognition and alleviate the misdiagnosis of abnormal rhythms.