Towards Non-Contact Glucose Sensing in Aqueous Turbid Medium at ~1.1 Meters Distance

TL;DR

This study presents a non-contact diffuse reflectance method for glucose sensing at approximately 1.1 meters distance, demonstrating linear response in turbid aqueous media with high accuracy, potentially enabling remote glucose monitoring.

Contribution

It introduces a novel non-contact diffuse reflectance technique at ~1.1 meters for glucose sensing, with experimental validation and high correlation in turbid media, advancing remote health monitoring capabilities.

Findings

Linear relationship between diffuse reflectance and glucose concentration.

High R^2 values (>0.987) indicating strong model fit.

Effective detection of glucose up to 17.8 mg/dL in turbid media.

Abstract

This work demonstrates a non-contact diffuse reflectance approach with a working distance of ~1.1 meters for the potential of glucose sensing. Non-contact diffuse reflectance over 1.1-1.3 micrometers was developed according to a center-illumination-area-detection (CIAD) geometry. The modeled response of diffuse reflectance in the CIAD geometry was examined with phantoms by altering independently the size of the collection geometry and the reduced scattering and absorption properties of the medium. When applied to aqueous turbid medium containing glucose control solutions with the cumulative volume varying over three orders of magnitude, a linear relationship expected for the diffuse reflectance as a function of the medium absorption/reduced-scattering property was observed for four conditions of the glucose-medium composition that differed either in the effective glucose concentration or the host medium scattering property. The cumulation of glucose up to 17.8mg/dL and 8.9mg/dL in the host medium having the same optical properties resulted in linear regression slopes of 0.0032 and 0.0030, respectively. The cumulation of the glucose up to 17.8mg/dL in an aqueous host medium that differed two folds in the reduced scattering property caused the linear regression slope to differ between 0.0032 and 0.0019. The R^2 values of all cases were all greater than 0.987. A provisional patent (US#63/053,004) has been obtained for this work.