Remote Blood Oxygen Estimation From Videos Using Neural Networks

Abstract

<div>Blood oxygen saturation (SpO<sub>2</sub>) is an essential indicator of respiratory functionality and is receiving increasing attention during the COVID-19 pandemic. Clinical findings show that it is possible for COVID-19 patients to have significantly low SpO<sub>2</sub> before any obvious symptoms. The prevalence of cameras has motivated researchers to investigate methods for monitoring SpO<sub>2 </sub>using videos. Most prior schemes involving smartphones are contact-based: They require a fingertip to cover the phone's camera and the nearby light source to capture re-emitted light from the illuminated tissue. In this paper, we propose the first convolutional neural network based noncontact SpO<sub>2</sub> estimation scheme using smartphone cameras. The scheme analyzes the videos of a participant's hand for physiological sensing, which is convenient and comfortable, and can protect their privacy and allow for keeping face masks on.</div><div>We design our neural network architectures inspired by the optophysiological models for SpO<sub>2</sub> measurement and demonstrate the explainability by visualizing the weights for channel combination. Our proposed models outperform the state-of-the-art model that is designed for contact-based SpO<sub>2</sub> measurement, showing the potential of our proposed method to contribute to public health. We also analyze the impact of skin type and the side of a hand on SpO<sub>2</sub> estimation performance.</div>

Remote Photoplethysmography (rPPG): A State-of-the-Art Review

“…Previous research studies have used a variety of different equipment and set-up to capture data from participants to obtain vital signs. Thermal camera [67], [68], [69], [70], Charge-Coupled Device (CCD) camera [71], [72], other affordable web cameras or those built-in laptops [73], [74], [75], [76], [77], [78], [79], [80], [81], MS Kinect V2 [82], [83], [66], Kinect Azure [84], GoPro camera with a drone [85] and smartphone [86], [87], [88], [89], [90], [91], [92], [93], [94] cameras have been previously used to obtain a person’s vital signs. Table V and Table IV show different studies conducted with various equipment and the vital sign under observation.…”

“…However, using palms under a camera is not very practical as people in real-life would be required to keep their hands still under a camera. In addition to that, hands are also a less exposed part of the body compared to that of the face, which would make it difficult to deploy it in a real-life scenario [86], [238].…”