Compressed sensing framework for BCG signals based on the optical fiber sensor.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Ballistocardiography (BCG) signals offer non-invasive cardiac monitoring.
• Traditional BCG data acquisition can be bandwidth-intensive.
• Compressed Sensing (CS) and deep learning (DL) show promise for efficient signal processing.

Purpose of the Study:

• To develop and evaluate a novel CS framework for BCG signal acquisition and reconstruction.
• To integrate an optical fiber sensor with a CS module and a DL algorithm.
• To assess the framework's performance across various compression ratios (CRs).

Main Methods:

• An optical fiber sensor-based system was designed to collect BCG data.
• A CS module compressed the BCG signals at the sensing end.
• An end-to-end DL algorithm reconstructed the compressed data.
• Performance was compared against traditional CS algorithms and a DL reference method.

Main Results:

• The proposed CS framework successfully reconstructed BCG signals from 50% to 95% CR.
• The framework outperformed existing methods, especially at high CRs.
• Mean Absolute Error (MAE) for heart rate (HR) estimation remained below 1 bpm at CRs below 95%.

Conclusions:

• The developed CS framework enables efficient BCG signal acquisition and reconstruction.
• The system demonstrates high accuracy and robustness for remote heart rate monitoring.
• The framework has significant potential for integration into Internet of Medical Things (IoMT) systems for healthcare applications.