Impact of Different Skin Penetration Depths of Red and Green Wavelengths on Camera-based SpO2 Measurement

Remote camera-based estimation of blood oxygen saturation (SpO(2)) using visible lights has been studied recently, typically for red (660 nm) and green (550 nm) wavelengths. This paper investigates the impact of different skin penetration depths of red and green wavelengths on the SpO(2) estimation based on mathematical modeling and experiments, where the SpO(2)-calibritability between two illumination setups, narrow-band red/green and narrow-band red/infrared, are statistically compared using the "ratio-of-ratios" method. The results show that the performance of the setup using red/green is less consistent among 17 volunteers than the setup using red/infrared, and larger SpO(2) disparity between different skin regions (by SpO(2) imaging) have been found for individuals in the red/green wavelengths setup. The use of visible light (red and green) may impose a risk of SpO(2) calibration due to the different skin penetration depths of these two wavelengths.