Related Experiment Videos
A Deep-Learning Approach to ECG Classification Based on Adversarial Domain Adaptation
Lisha Niu1, Chao Chen1, Hui Liu1
1Shandong Artificial Intelligence Institute, Qilu University of Technology (Shandong Academy of Sciences), Jinan 250353, China.
Insights
A new deep learning method improves electrocardiogram (ECG) classification accuracy by using adversarial domain adaptation. This approach addresses limited training data and variations in ECG signals, achieving 92.3% accuracy.
Area of Science:
- Medical Informatics
- Artificial Intelligence
- Biomedical Signal Processing
Background:
- Cardiovascular disease poses a significant global health threat, necessitating efficient diagnostic tools.
- Current diagnostic methods for cardiovascular disease, particularly those relying on electrocardiogram (ECG) signals, face limitations due to scarce medical resources and the need for expert interpretation.
- Computer-aided diagnosis systems offer a promising solution to overcome these limitations by automating ECG analysis.
Purpose of the Study:
- To develop a novel deep-learning method for accurate ECG classification, specifically addressing challenges of insufficient labeled training samples and cross-domain data distribution discrepancies.
- To enhance the classification accuracy of ECG signals from different distributions caused by individual variations.
Main Methods:
- A deep-learning framework incorporating three modules: multi-scale feature extraction (F), domain discrimination (D), and classification (C).
- Module F utilizes parallel convolution blocks for comprehensive feature extraction.
- Module D employs convolutional blocks and a fully connected layer to address low model layers and feature abstraction.
- Module C concatenates time and deep-learning extracted features for enhanced diversity.
Main Results:
- The proposed method achieved a classification accuracy of 92.3% on experimental ECG signals.
- The adversarial domain adaptation technique effectively mitigated issues related to insufficient labeled data and varied data distributions.
- Experimental validation confirmed the method's effectiveness in cross-domain ECG signal classification.
Conclusions:
- The novel deep-learning approach based on adversarial domain adaptation significantly improves ECG classification accuracy.
- This method offers a robust solution for automated cardiovascular disease diagnosis, particularly in resource-limited settings.
- The enhanced feature diversity and cross-domain adaptability make the system suitable for real-world clinical applications.
Abstract:
Cardiovascular disease has become one of the main diseases threatening human life and health. This disease is very common and troublesome, and the existing medical resources are scarce, so it is necessary to use a computer-aided automatic diagnosis to overcome these limitations. A computer-aided diagnostic system can automatically diagnose through an electrocardiogram (ECG) signal. This paper proposes a novel deep-learning method for ECG classification based on adversarial domain adaptation, which solves the problem of insufficient-labeled training samples, improves the phenomenon of different data distribution caused by individual differences, and enhances the classification accuracy of cross-domain ECG signals with different data distributions. The proposed method includes three modules: multi-scale feature extraction F, domain discrimination D, and classification C. The module F, constitutive of three different parallel convolution blocks, is constructed to increase the breadth of features extracted from this module. The module D is composed of three convolutional blocks and a fully connected layer, which is to solve the problem of low model layers and low-feature abstraction. In the module C, the time features and the deep-learning extraction features are concatenated on the fully connected layer to enhance feature diversity. The effectiveness of the proposed method is verified by experiments, and the classification accuracy of the experimental electrical signals reaches 92.3%.
