Abstract
Glucose sensing is pursued extensively in biomedical research and clinical practice for assessment of the carbohydrate and fat metabolism as well as in the context of an array of disorders, including diabetes, morbid obesity, and cancer. Currently used methods for real-time glucose measurements are invasive and require access to body fluids, with novel tools and methods for non-invasive sensing of the glucose levels highly desired. In this study, we introduce a near-infrared (NIR) optoacoustic spectrometer for sensing physiological concentrations of glucose within aqueous media and describe the glucose spectra within 850-1,900 nm and various concentration ranges. We apply the ratiometric and dictionary learning methods with a training set of data and validate their utility for glucose concentration measurements with optoacoustics in the probe dataset. We demonstrate the superior signal-to-noise ratio (factor of ~3.9) achieved with dictionary learning over the ratiometric approach across the wide glucose concentration range. Our data show a linear relationship between the optoacoustic signal intensity and physiological glucose concentration, in line with the results of optical spectroscopy. Thus, the feasibility of detecting physiological glucose concentrations using NIR optoacoustic spectroscopy is demonstrated, enabling the sensing glucose with ±10 mg/dl precision.