Enhancing visual seismocardiography in noisy environments with adaptive bidirectional filtering for Cardiac Health...

Area of Science:

• Biomedical Engineering
• Cardiovascular Monitoring
• Signal Processing

Background:

• Wearable sensors are transforming cardiac health monitoring through non-invasive Seismocardiography (SCG).
• Motion artifacts, particularly from walking, significantly degrade SCG signal quality and limit clinical applications.
• Existing methods struggle to effectively remove noise while preserving essential cardiac information.

Purpose of the Study:

• To introduce and evaluate an innovative Adaptive Bidirectional Filtering (ABF) technique for motion artifact reduction in SCG signals.
• To demonstrate ABF's superior performance in refining SCG signals compared to conventional methods.
• To enable reliable cardiac monitoring in real-world, noisy environments using wearable sensors.

Main Methods:

• The ABF technique employs a two-stage process involving Redundant Multi-Scale Wavelet Decomposition (RMWD) and adaptive bidirectional filtering.
• RMWD identifies and isolates heart-associated signals and frequencies.
• Adaptive bidirectional filtering utilizes Time-Frequency masking and non-negative matrix decomposition for noise elimination and spatial correlation, incorporating motion-adapted recursive structures using vertical accelerometer data.

Main Results:

• ABF achieved an r-squared value of 0.95 for heart rate estimation accuracy at -20 dB SNR, outperforming baseline values (0.1-0.85).
• Effective motion artifact reduction was demonstrated even at -22 dB SNR, eliminating the need for Electrocardiography (ECG) inputs.
• The method filters up to 97% of motion-related noise in SCG signals from implantable devices and enhances ECG filtering and beat detection.

Conclusions:

• The Adaptive Bidirectional Filtering (ABF) technique offers a robust solution for mitigating motion artifacts in Seismocardiography (SCG).
• ABF significantly enhances the accuracy and reliability of cardiac monitoring using wearable sensors, even in challenging, noisy conditions.
• This advancement is expected to be integral to routine patient monitoring, improving diagnostic capabilities and reducing reliance on supplementary sensors like ECG.