Skip navigation
Please use this identifier to cite or link to this item:
Title: Noninvasive glucose monitoring: New opportunities opened by mid-infrared quantum cascade laser spectroscopy
Authors: Werth, Alexandra Margot
Advisors: Gmachl, Claire F
Contributors: Electrical Engineering Department
Keywords: Glucose
Machine learning
Medical device
Quantum cascade laser
Subjects: Optics
Issue Date: 2019
Publisher: Princeton, NJ : Princeton University
Abstract: Diabetes is a growing problem throughout the world, and for many people with diabetes it is imperative to closely monitor their glucose level throughout the day. Diabetes creates large swings in the glucose concentration in blood posing imminent dangers such as seizures, unconsciousness, and if not treated, death as well as long term complications such as heart and blood vessel disease, kidney failure, and nerve damage. Fortunately, administration of medication or insulin can mitigate these high and low swings of blood sugar. However, in order to accurately administer these medications one needs to closely monitor their blood sugar levels. Current methods of monitoring blood glucose concentrations involve a painful finger prick or a monitor which needs to be embedded under the skin. The development of a noninvasive, pain-free, continuous glucose monitor would therefore improve the quality of life and health of hundreds of millions of people suffering with diabetes. Mid-infrared quantum cascade laser spectroscopy has the potential to satisfy these needs. We have developed a noninvasive glucose sensor based on mid-infrared quantum cascade laser spectroscopy which operates in the 8 - 10 μm wavelengths region. This wavelength range contains unique spectral absorption features of glucose, particularly the C-O stretching mode at 9.5 μm, which are much stronger than that of other carbohydrates in this range. The light penetrates into the dermis layer of the skin where it is absorbed by the glucose molecules in the interstitial fluid. The light is then backscattered off of the collagen fibers and collected using the integrating sphere and mercury cadmium telluride (MCT) detector. We analyze a series of collected spectra using multivariate algorithms, including principal component (PC) regression and partial least square (PLS) regression, to determine the wavelengths that correspond to highest variance. We consistently see a strong correlation with the predicted PC loading vector of the spectra and the known glucose absorption spectrum. In a small-scale usability study with the most recent generation of the sensor, we have predicted in vivo glucose concentrations of single subjects with over 80% accuracy with 66% of the results falling within ±15% error. After implementation of a finger demobilizer clip and pressure sensor on the sample port of the integrating sphere we were able to increase the accuracy of the predicted glucose concentrations to over 90% with 91% of the results falling within ±15% error for a single subject. The latter results are just shy of clinical accuracy, 95% of results falling within ±15% error, as determined by the FDA. Lastly, we propose a new experiment to test the feasibility of in vivo blood alcohol monitoring through the skin using the sensor we have developed. Blood alcohol monitoring has its own set of useful applications; however, beyond these independent applications, adapting our sensor for blood alcohol monitoring could improve the accuracy of the glucose sensor. The spectral features of ethanol closely overlap with features in the glucose spectrum within the 8 - 10 μm window. Not only are the spectra similar, but the concentrations of ethanol and glucose in the blood can be very similar implying that it is important to consider blood alcohol concentrations when optically monitoring glucose concentrations in the interstitial fluid.
Alternate format: The Mudd Manuscript Library retains one bound copy of each dissertation. Search for these copies in the library's main catalog:
Type of Material: Academic dissertations (Ph.D.)
Language: en
Appears in Collections:Electrical Engineering

Files in This Item:
File Description SizeFormat 
Werth_princeton_0181D_13018.pdf22.58 MBAdobe PDFView/Download

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.