Modeling on the Feasibility of Camera-Based Blood Glucose Measurement

Yiyin Wang, Wenjin Wang, Mark van Gastel, Gerard de Haan; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 0-0

Monitoring of blood glucose levels is crucial for diabetics to manage their lives. However, the current gold-standard requires taking invasive blood samples, which is painful and can lead to infections. In this paper, we investigate the feasibility of using a regular camera (with silicon image sensors) to estimate the blood glucose levels remotely as claimed by recent studies. The physiological challenge is the small volume fraction and low absorption of glucose in the human body as compared to other absorbers. The glucose-induced variations in light intensity from both the non-pulsating and the pulsating part of the reflected optical signal are modeled in the visible to near-infrared wavelength range. The simulation results suggest that it is unlikely to detect the blood glucose based on either the DC or AC component of skin reflected light. The optical responses caused by glucose changes are minor as compared to other physiological factors (e.g. skin temperature, SaO2, water concentration). This, combined with the coarse sampling of the light spectrum by regular cameras render the measurement infeasible.