A 2.2 nW Analog Electrocardiogram Processor Based on Stochastic Resonance Achieving a 99.94% QRS Complex Detection...

Area of Science:

• Biomedical Engineering
• Analog Signal Processing
• Wearable Health Technology

Background:

• Electrocardiogram (ECG) signal processing is crucial for diagnosing cardiac conditions.
• Existing ultra-low power ECG processors often struggle with real-time detection and noise suppression.
• The integration of stochastic resonance offers a novel approach to enhance signal detection in noisy environments.

Purpose of the Study:

• To develop an ultra-low power ECG processor for real-time QRS-wave detection.
• To implement advanced noise suppression and signal enhancement techniques.
• To validate the processor's performance against established ECG databases and compare it with digital algorithms.

Main Methods:

• Designed and implemented an ultra-low power ECG processor using current-mode analog signal processing in 65 nm CMOS technology.
• Employed a linear filter for out-of-band noise suppression and a nonlinear filter for in-band noise suppression and QRS-wave enhancement via stochastic resonance.
• Utilized a constant threshold detector for QRS-wave identification on processed ECG signals.

Main Results:

• Achieved an average F1 score of 99.88% on the MIT-BIH Arrhythmia database, surpassing previous ultra-low power ECG processors.
• Demonstrated superior detection performance on noisy ECG recordings from MIT-BIH NST and TELE databases compared to most digital algorithms.
• The processor boasts a minimal footprint (0.08 mm²) and extremely low power dissipation (2.2 nW at 1V).

Conclusions:

• The developed ultra-low power ECG processor enables real-time QRS-wave detection with exceptional accuracy and energy efficiency.
• The integration of stochastic resonance in an analog processor represents a significant advancement in wearable ECG monitoring.
• This processor is the first of its kind, offering validated real-time performance with stochastic resonance for noisy ECG data.