Accuracy enhancement in reflective pulse oximetry by considering wavelength-dependent pathlengths

Area of Science:

• Biomedical Engineering
• Physiological Measurement
• Wearable Technology

Background:

• Transmissive photoplethysmography (tPPG) is standard for SpO2, but reflective PPG (rPPG) in wearables lacks acceptance due to patient-dependent pathlengths.
• Existing rPPG methods, like the 'Ratio of Modulation' (R), require calibration, limiting their clinical utility.

Purpose of the Study:

• To develop a calibration-free algorithm for accurate SpO2 measurement using rPPG.
• To introduce and validate a 'pathlength ratio' (β) to compensate for patient-specific variations in rPPG signals.

Main Methods:

• Derived a pathlength ratio (β) from an analytical model of rPPG signals.
• Applied β to a calibration-free algorithm and validated using data from a human hypoxia study with Blood Gas Analysis as reference.
• Analyzed the natural logarithm of red and infrared PPG signals in conjunction with β.

Main Results:

• The proposed method significantly reduced the root-mean-square error (RMSE) in SpO2 measurement to an average of 1%.
• The pathlength ratio (β) demonstrated effectiveness across a clinical SpO2 range of 70% to 100%.

Conclusions:

• The calibration-free rPPG method, incorporating the pathlength ratio (β), can achieve SpO2 measurements with accuracy below the 2% clinical threshold.
• This approach extends the applicability of tPPG-based calibration-free algorithms to rPPG devices, reducing estimation parameters.