EFFECTS OF SKIN TONE AND AMBIENT LIGHTING ON ACCURACY OF CONTACTLESS MEASUREMENT OF HEART RATE AND PULSE TRANSIT TIME: AN INVESTIGATION TOWARD IMAGE-BASED BLOOD PRESSURE ESTIMATION

Abstract

Abstract Objective: Image-based photoplethysmography (iPPG) using video of the skin offers potential for contactless blood pressure (BP) estimation from the pulse transit time (PTT). Optical properties of human skin and ambient lighting are likely to affect accuracy of image-based PTT (iPPT) and are investigated here. Design and method: Twelve healthy volunteers (6 female, 27 ± 6 years) with a range of skin tones (Fitzpatrick skin scores of 5 to 27) were recruited. For each subject, the experiment was performed in standard fluorescent room lighting (100 lux) and repeated at five other levels of light intensity (40–310 lux). Two-minute videos were acquired from each subject's face and palm of the left hand; simultaneous electrocardiogram (ECG) and non-invasive finger BP were also measured. In each frame, two regions of interest (ROIs) were selected from the subject's face and hand, and the average of green colour intensity for each ROI were calculated as iPPG. The iPPG signal was band-pass filtered between 0.66–3 Hz, (corresponding to a 40–180 bpm heart rate (HR) range). HR was calculated as the dominant peak in the Fourier spectrum of filtered iPPG. Beat-to-beat values of iPTT were calculated between the corresponding feet of the iPPG pulsations. Results: The absolute error between average HR (errHR) determined from ECG and that calculated from iPPG in both ROIs was significantly correlated with room lumen (RL) and skin score (SC) (errHR = – 0.08 × RL + 0.67 × SC + 12, p < 0.05). The minimum HR error was in standard room lighting (errHR = 3.9 ± 8.9 bpm). Using average iPTT in room lighting as the reference, the percentage error (iPTT%err) between reference iPTT and iPTT in other lighting conditions significantly decreased with lower lighting conditions but was not significantly changed with skin tone score (iPTT%err = -0.09 × RL + 55, p < 0.05). Conclusions: Skin tone and ambient lighting intensity affects accuracy of contactless HR measurement. Only light intensity affects accuracy of iPTT measurement. It is likely that errors in contactless BP estimation are minimised in standard room lighting conditions but increase in dimmer conditions.