Embedded hardware-based adaptive filtering for noise reduction in bioimpedance data.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Wearable Technology

Background:

• Increasing use of wearable devices for health monitoring necessitates robust bioimpedance data acquisition.
• Integrated circuits like AD5933 enable affordable bioimpedance solutions, but noise reduction remains a challenge.
• Traditional digital filters often require manual tuning, limiting their adaptability to varying bioimpedance signals.

Purpose of the Study:

• To develop an automated digital filter for effective noise reduction in bioimpedance data.
• To enable adaptive filter coefficient adjustment without prior signal knowledge.
• To validate the method's performance and applicability in portable bioimpedance systems.

Main Methods:

• An automated digital filter design based on minimizing the smoothness difference between consecutive filtered data points.
• Implementation on a microcontroller for real-time processing.
• Testing with synthetic and experimental bioimpedance, electromyography (EMG), and respiration data.

Main Results:

• Achieved up to 8 dB signal-to-noise ratio improvement for noise levels up to 2%.
• Demonstrated low power consumption (<11 mW) and fast execution time (<185 ms).
• Successfully filtered diverse bioimpedance signals, including EMG and respiration.

Conclusions:

• The proposed automated filter offers efficient and adaptive noise reduction for bioimpedance data.
• The method is suitable for low-power, portable, and wearable bioimpedance applications.
• The filter's versatility extends to various biological signals, enhancing its practical utility.