ECG beat classification with fractional order differentiator and machine learning techniques

Area of Science:

• Biomedical Engineering
• Cardiology
• Signal Processing

Background:

• Manual electrocardiogram (ECG) analysis is laborious and prone to errors.
• Automated ECG analysis is crucial for early cardiovascular disease detection, especially with irregular heartbeats.
• Accurate identification of abnormal heartbeats is essential for timely diagnosis and treatment.

Purpose of the Study:

• To develop a novel, accurate approach for automated ECG beat classification.
• To integrate fractional order differentiation, dual-tree complex wavelet transform (DTCWT) features, and machine learning (ML) for enhanced ECG analysis.
• To improve the diagnostic accuracy of ECG interpretation for better clinical outcomes.

Main Methods:

• R-peak detection was performed using a fractional order differentiator.
• Feature extraction was conducted using the dual-tree complex wavelet transform (DTCWT).
• Various machine learning (ML) classifiers were employed for ECG beat classification, including Random Forest.

Main Results:

• The proposed methodology demonstrated superior performance on the MIT-BIH Arrhythmia Database.
• The Random Forest classifier achieved high accuracy (96.82%), sensitivity (96.83%), specificity (97.02%), PPV (96.89%), and F1-score (96.85%).
• The integrated approach effectively handles signal irregularity and non-stationarity in ECG data.

Conclusions:

• The proposed method significantly enhances the accuracy of ECG beat classification.
• This approach contributes to more reliable early detection of cardiovascular diseases.
• Improved ECG analysis accuracy can lead to better clinical decision-making and patient outcomes.