Abstract
Inaccurate oxygen saturation (SpO(2)) measurements using photoplethysmography (PPG) based pulse oximeters on dark skin tones, were a direct cause of increased mortality during the Covid pandemic (2019-2022) and continue to negatively impact routine clinical monitoring. Inaccuracies are commonly interpreted as being caused by rich pigmentation in dark skin tones. A multispectral optoelectronic sensor (mSOS) with bespoke electronics has been created to overcome skin tone related difficulties of capturing peripheral blood pulsatile variations in deoxyhemoglobin (HHb) in red visible spectrum (>600 nm), and to improve the detection of oxyhaemoglobin (HbO(2)) in infrared spectrum (>850 nm). The study demonstrates that a better understanding of the impact of skin tone on complex tissue optics is required to address pulse oximeter performance difficulties arising from skin pigmentation. The study evaluates the relationship between signal quality and skin tones with spectral illuminations at 515 nm, 601 nm, 631 nm and 940 nm. An approved hypoxia protocol with twelve subjects across three different skin tone groups (Individual Typology Angles (ITA): 50(°) - 41(°), Monk Skin Tones: 2, 3 (Group I), ITA: 23(°) - 1(°); MST:4, 5 (Group II), and ITA: -31° - -66(°), MST:8, 9 and 10 (Group III)) was implemented with the mSOS attached to the back of subject's wrist, including six subjects of Group III. The results show the mSOS is capable of detecting peripheral blood perfusion variations with equivalent performance across three skin tone groups for all four wavelengths and throughout the oxygen desaturation range from 100% to 70%. Performance was maintained for all twelve subjects irrespective of skin tone as judged by signal to noise ratio (SNR) with the range of 19.19 - 25.00 and the standard deviation in the range of 1.42 - 5.60. Signal quality (SNR) remains consistently high across all the illumination wavelengths for all subjects, at all desaturation levels (100 - 70%). Upcoming studies will investigate SpO(2) calibration against SaO(2) gold standard reference with a larger subject cohort.