CLINet: A novel deep learning network for ECG signal classification

Area of Science:

• Artificial Intelligence in Medicine
• Cardiology
• Biomedical Signal Processing

Background:

• Cardiac arrhythmia poses a significant health risk, necessitating efficient diagnostic tools.
• Current diagnostic methods can be time-consuming and require specialized expertise.
• Automated analysis of electrocardiogram (ECG) signals offers a promising avenue for improved arrhythmia detection.

Purpose of the Study:

• To introduce CLINet, a novel deep learning network for automated ECG signal classification.
• To evaluate the performance of CLINet in identifying cardiac arrhythmia disorders.
• To demonstrate the feasibility of deploying an accurate arrhythmia detection model on resource-constrained devices.

Main Methods:

• Development of CLINet, a deep learning network integrating convolution, LSTM, and involution layers.
• Utilizing multiple, large-size kernels in convolution and involution layers for multi-scale feature learning.
• Designing CLINet for minimal pre-processing requirements and adaptability to ECGs of varying lengths.

Main Results:

• CLINet achieved high accuracy: 99.90% on the ICCAD dataset and 99.94% on the MIT-BIH dataset.
• The model boasts a compact size with only 297K parameters, facilitating integration into smart devices.
• No complex pre-processing steps were needed, simplifying the diagnostic workflow.

Conclusions:

• CLINet demonstrates exceptional accuracy and efficiency in classifying ECG signals for cardiac arrhythmia detection.
• The network's lightweight design makes it ideal for integration into wearable technology for continuous patient monitoring.
• This deep learning approach has the potential to significantly improve the early diagnosis and management of life-threatening arrhythmia conditions.