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

Abstract

Objective: Global (inter-patient) ECG classification for arrhythmia detection over Electrocardiogram (ECG) signal is a challenging task for both humans and machines. Automating this process with utmost accuracy is, therefore, highly desirable due to the advent of wearable ECG sensors. However, even with numerous deep learning approaches proposed recently, there is still a notable gap in the performance of global and patientspecific ECG classification performance. Methods: In this study, we propose a novel approach for inter-patient ECG classification using a compact 1D Self-ONN by exploiting morphological and timing information in heart cycles. We used 1D Self-ONN layers to automatically learn morphological representations from ECG data, enabling us to capture the shape of the ECG waveform around the R peaks. We further inject temporal features based on RR interval for timing characterization. The classification layers can thus benefit from both temporal and learned features for the final arrhythmia classification. Results: Using the MIT-BIH arrhythmia benchmark database, the proposed method achieves the highest classification performance ever achieved, i.e., 99.21% precision, 99.10% recall, and 99.15% F1-score for normal (N) segments; 82.19% precision, 82.50% recall, and 82.34% F1-score for the supra-ventricular ectopic beat (SVEBs); and finally, 94.41% precision, 96.10% recall, and 95.2% F1-score for the ventricular-ectopic beats (VEBs). Significance: As a pioneer application, the results show that compact and shallow 1D Self-ONNs with the feature injection can surpass all state-ofthe-art deep models with a significant margin and with minimal computational complexity. Conclusion: This study has demonstrated that using a compact and superior network model, a global ECG classification can still be achieved with an elegant performance level even when no patient-specific information is used.

A robust multiple heartbeats classification with weight-based loss based on convolutional neural network and bidirectional long short-term memory

“…The best performances for each evaluation index are shown in bold font. It is worth noting that the algorithm proposed by Zahid et al ( Zahid et al, 2022 ) showed the best specificity 99.83% for N heartbeat type, and Sellami et al ( Sellami and Hwang, 2019 ) presented the higher accuracy 99.99% and specificity 89.54% for S heartbeat type. However, the specificity and accuracy are not the reasonable measure indexes in imbalanced dataset as shown in Table 8 .…”

“…Prabhakararao et al ( Prabhakararao and Dandapat, 2021 ) designed a classifier based multiple scale-dependent deep convolutional neural networks with different receptive fields for arrhythmia classification, the model showed impressive performance (averaged 84.5% F 1 score on PTBXL-2020 dataset and 88.3% F 1 score on CinC-2017 dataset) and generalization ability, and then made it suitable for arrhythmia monitoring applications. Zahid et al ( Zahid et al, 2022 ) used MIT-BIH arrhythmia dataset and proposed a novel model combined temporal feature based on RR interval and learned features to classify arrhythmia, the F 1 score is 99.15% for super-ventricular ectopic beats and 95.2% for ventricular-ectopic beats. Khatibi et al ( Khatibi and Rabinezhadsadatmahaleh, 2019 ) proposed a novel feature engineering method based on deep learning and K-NNs showing a good performance to classify heartbeat.…”

Contextual attention U-net: a deep learning-based model for decoding brain characteristics of smokers

“…Since CAT is to identify and represent smoking-related brain features, we use three classical ML-based methods (e.g., KNN [7], RF [8], and SVM [9]) and seven well-known DL-based approaches (e.g., Xception [17], VGG 19 [18], VGG 16 [19], ResNet 18 [20], Mobile [21], Shuffle [22], and DenseNet [23]) as baselines for comparisons. For the five datasets, the training model of each baseline strategy is re-trained, to provide optimal results for fair comparisons.…”

An improved SR-LSTM network for atrial fibrillation and atrial flutter signals discrimination

“…As for MARD, it consists of 48 2-leads recordings and each recording is sampled with a frequency of 360 Hz with the duration of 30 min [32]. Further, we will de-110 sample the recordings of MARD into 250 Hz for data consistency dividing them into 2 s fragments [33,34], implement denoising by wavelet transform strategy and carry out Z-score normalization on filtered ECG fragments for accelerating convergence [45]. Table 1 shows details of ECG fragments in this research.…”