Smartphone camera oximetry in an induced hypoxemia study

Abstract

Hypoxemia, a medical condition that occurs when the blood is not carrying enough oxygen to adequately supply the tissues, is a leading indicator for dangerous complications of respiratory diseases like asthma, COPD, and COVID-19. While purpose-built pulse oximeters can provide accurate blood-oxygen saturation (SpO2) readings that allow for diagnosis of hypoxemia, enabling this capability in unmodified smartphone cameras via a software update could give more people access to important information about their health. Towards this goal, we performed the first clinical development validation on a smartphone camera-based SpO2 sensing system using a varied fraction of inspired oxygen (FiO2) protocol, creating a clinically relevant validation dataset for solely smartphone-based contact PPG methods on a wider range of SpO2 values (70–100%) than prior studies (85–100%). We built a deep learning model using this data to demonstrate an overall MAE = 5.00% SpO2 while identifying positive cases of low SpO2 < 90% with 81% sensitivity and 79% specificity. We also provide the data in open-source format, so that others may build on this work.

A calibration method for smartphone camera photophlethysmography

“…This can largely be credited to the use of data-driven models. Several works have used data-driven models to successfully measure blood oxygenation using the smartphone camera with the flash LED for SpO2 measurement, achieving an MAE of 5% or less (8,9). For anemia screening via HgB measurement, Wang et al (10) leveraged a Support Vector Machine (SVM) to train a model utilizing various combinations of amplitude and ratios achieving a rank order correlation of 0.82.…”

“…Photoplethysmography (PPG) is a widely used measurement technique in the medical field that involves illuminating a vascularized tissue and tracking the reflected light over time. This pivotal measurement is adept at capturing an array of physiologically meaningful information, such as heart rate (1)(2)(3)(4)(5), heart rate variability (6,7), blood oxygenation (5,8,9), hemoglobin concentration (10,11), vascular aging (12), blood pressure (5,(13)(14)(15), diabetes (16), and respiratory rate (17). A PPG measurement is realized with a light source and a photo sensor pair.…”

“…Most prior works largely depended on post-processing after allowing the complex camera system to automatically capture a video, with autoexposure, auto-white balancing, and video compression algorithms altering the raw signal in non-linear ways (4,6,(20)(21)(22)(23)(24)(25)(26)(27)(28). Only a handful of works have addressed changing exposure by disabling auto-exposure (29)(30)(31); calculating absorption frameby-frame in an online manner to avoid compression; attempting to control white balance by designating a preset, most commonly an incandescent setting to emphasize blue and green channels (32)(33)(34); or opting to control each color channels with individual gain settings (8,11). Although these works hint at the right direction in addressing the issues, our investigation finds that these solutions are inadequate, and in fact, overlook some of the most crucial camera parameters affecting cPPG measurements.…”