Mobile phone camera-based SpO2 measurements using broadband light and colored paper filters (Conference Presentation)

The mobile health field has given rise to a surge of point-of-care diagnostic attachments for mobile phones. These attachments, however, are limited in adoption in low-resource settings due to initial acquisition and subsequent maintenance cost challenges. Point-of-care devices that require no or minimum attachment can make a great impact to the accessibility of such devices in resource-poor regions. In this abstract, we report a simulation study to demonstrate the feasibility of using an ultra-low-cost color-paper filter and a mobile phone to perform broadband pulse oximetry. We run a series of GPU-based Monte Carlo simulations using a previously segmented 7T MRI scan of a finger 3D model. We sweep the optical properties of the finger tissues between the wavelengh band of 400-800 nm with a 1 nm increment, with intensity based on the measured spectrum of an iPhone 8’s LED. We also measured the transmission spectra from paper filters of various colors, which we used to further alter the light source spectrum. Using a discretized photoplethysmogram (PPG) signal, we simulate a 60 bpm oscillation optical measurements due to an up to 15% volume changes of the finger arterioles. Simulations were repeated for various peripheral blood oxygen levels (SpO2). Finally, we estimate the SpO2 using the simulated PPG signals using the Ratio of Ratios (RR) method. We evaluate the performance of different color paper filters by comparing 1) total optical signal intensity, 2) maximum magnitude of the RR signal variations and 3) the correlation of the computed and assumed SpO2 values. We found that the purple-colored filter produced the highest RR signal variations and the cyan-colored paper resulted in the largest SpO2 changes in the tested range.