Glucose sensing using near infrared spectroscopy

Davison, Nicholas and Zhuang, Qiandong and Gaffney, Christopher and Kerns, Jemma (2024) Glucose sensing using near infrared spectroscopy. PhD thesis, Lancaster University.

Text (2024DavisonPhD)
2024davisonphd.pdf - Published Version
Restricted to Repository staff only until 18 July 2029.
Download (14MB)

Abstract

Absorption spectroscopy in the Near Infrared (NIR) region is a promising technology for non-invasive glucose sensing. Regular glucose monitoring is required to maintain glycaemic control in patients with diabetes, however, current glucose sensing methods, mostly based on needle-prick meters or subcutaneous sensors, are invasive and unpleasant for patients. NIR offers the advantages of strong tissue penetration, a well-developed detector technology platform and a method that measures the glucose absorption by infrared light. However, the accuracy of any NIR spectroscopy technique depends on sensor design factors including the choice of detector, light source, sample thickness, data pre-processing method or measurement geometry, which can be investigated using a lab-based spectroscopy setup. This study aimed to develop an NIR spectroscopy system and then investigate the spectra of glucose in solution, intralipid and bovine blood in both transmittance and reflectance modes. By collecting spectra of samples with a fixed glucose concentration difference and applying Partial Least Squares regression, the different techniques were evaluated using the Standard Error of Prediction (SEP). In aqueous solution, the most prominent glucose absorption band was the 1600-1700 nm first CH overtone and the limit of detection was 5 mmol/L. Intralipid spectra were dominated by glucose-dependant scattering, especially below 1300 nm, as expected by Mie scattering theory. Bovine blood spectra were dominated by water and haemoglobin, with increased absorbance across the entire spectral range at higher glucose concentrations. A type II InGaAs/GaAsSb superlattice detector was developed, aiming to target peak responsivity in the glucose combination band. The device had a dark current density of 0.137 A cm−2 and, when collecting spectra from glucose solutions with 100 mmol/L fixed step, had an SEP competitive to a commercial extended-InGaAs diode. This thesis demonstrates that NIR spectroscopy can to be used for glucose sensing and highlights the considerations when designing a non-invasive glucose-sensing device.