Advances in an In Vivo Noninvasive Glucose Sensing System: Using an Integrating Sphere, TE Cooled MCT Detector and Filtered Derivative Spectroscopy

Abstract

Each day millions of people with diabetes are required to prick their finger

multiple times in order to get the blood necessary to monitor their blood glucose levels. While invasive methods such as those involving the pricking of a finger are common and show high accuracy they bring with them pain and discomfort. In order to be viable as a commercial device, predictions of glucose concentrations must be within 20% of the actual concentration. Few non-invasive devices achieve this level of accuracy. Of the non-invasive methods that do, even fewer are considered inexpensive and none are as widely used as finger-pricking. In an attempt to change this status-quo, we have developed a non-invasive blood glucose level detection system using mid-infrared (mid- IR) spectroscopy. By decreasing the number of costly components of the noninvasive system and increasing the accuracy, it could become a widely-used technology. An integrating sphere was implemented in the system to collect more backscattered mid-IR light from the palm. Results show that the use of an integrating sphere improves the spectrum signal to noise ratio and consistency leading to the possibility for more accurate glucose level predictions. Additionally, a thermal-electrically cooled MCT detector has been used in place of the liquid nitrogen cooled detector. Employed signal processing techniques include Savitzky-Golay filtering and differentiation for derivative spectroscopy. The increase in SNR that occurs after SG differentiation heightens the possibility of using derivative spectroscopy for predictive analysis. Together, these changes to the system increase the viability of noninvasive glucose detection.