An Improved Sliding Window Area Method for T Wave Detection.

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• T wave detection is crucial for diagnosing myocardial ischemia and guiding clinical decisions.
• Current sliding area-based methods for T wave detection lack adaptive parameter settings.
• Accurate T wave detection is essential for daily mobile electrocardiogram (ECG) monitoring.

Purpose of the Study:

• To propose an improved sliding window area method for T wave detection with adaptive parameter settings.
• To enhance the accuracy and robustness of T wave detection in ECG signals.
• To provide a more reliable tool for mobile ECG monitoring and clinical diagnosis.

Main Methods:

• Utilized k-means clustering on the MIT QT database to establish relationships between RR intervals and T wave timing.
• Developed piecewise functions to delineate T wave onset and offset based on RR intervals.
• Employed grid search with 5-fold cross-validation to optimize parameters for the sliding window area method.

Main Results:

• Achieved significant improvements in F1 scores for T wave onset detection in the QT database (e.g., from 54.70% to 70.46% on one channel).
• Demonstrated enhanced F1 scores for T wave offset detection across multiple ECG databases.
• The improved algorithm consistently outperformed the traditional sliding window area method.

Conclusions:

• The proposed adaptive sliding window area method shows superior performance compared to traditional approaches.
• The enhanced T wave detection accuracy indicates potential utility in clinical ECG monitoring.
• This method offers a more robust and adaptive solution for real-time ECG analysis.