A universal, high-performance ECG signal processing engine to reduce clinical burden

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Electrocardiogram (ECG) signal conditioning is crucial for accurate noise removal and feature extraction.
• Effective ECG processing is vital for reliable diagnostic interpretation.

Purpose of the Study:

• To evaluate the performance of the FDA 510(k) cleared HeartKey Signal Conditioning and QRS peak detection algorithms.
• To assess algorithm efficacy across diverse ECG datasets, including public and proprietary sources.

Main Methods:

• Processed 751 raw ECG files from various use cases through the HeartKey signal processing engine.
• Utilized advanced filtering for noise reduction and QRS complex identification.
• Generated QRS detection statistics against annotated ECG files.

Main Results:

• Achieved high QRS detection accuracy (Se 98.90%, PPV 99.08%) on public databases (NSR, MIT-BIH, AHA, MIT-AF).
• Demonstrated robust performance (Se 93.26%, PPV 90.53%) on the challenging NST database.
• Showcased excellent performance with dry electrodes (Se 99.22%, PPV 99.00%) across eight databases, including motion-based recordings.

Conclusions:

• HeartKey exhibits robust signal conditioning and QRS detection across a wide array of ECG signals.
• The algorithm's consistent performance suggests it is a universal solution for diverse clinical and everyday applications without database-specific optimization.