Application of deep learning techniques for heartbeats detection using ECG signals-analysis and review

Area of Science:

• Cardiology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Electrocardiogram (ECG) data analysis is crucial for diagnosing cardiac arrhythmias.
• Deep learning (DL) models have emerged as powerful tools for automated ECG classification.
• Existing research employs diverse DL techniques, necessitating a comprehensive review.

Purpose of the Study:

• To provide a detailed examination of deep learning methods applied to ECG arrhythmia detection.
• To survey and analyze various DL approaches and their contributions to the field.
• To present findings from experimental studies using a large-scale ECG dataset.

Main Methods:

• Systematic survey of journal papers detailing DL methods for ECG analysis.
• Description and discussion of various DL models and experimental setups.
• Utilized a five-class ECG dataset comprising 100,022 heartbeats for model evaluation.

Main Results:

• Identified and categorized diverse deep learning techniques employed in ECG arrhythmia detection.
• Presented experimental results from applying constructed DL models to the five-class ECG dataset.
• Demonstrated the performance of different DL approaches in classifying ECG data.

Conclusions:

• Deep learning offers significant potential for accurate ECG arrhythmia classification.
• Further research is suggested to refine DL models and explore novel architectures for improved diagnostic capabilities.
• This review provides valuable insights for researchers and clinicians in the field of automated cardiac arrhythmia detection.