Photoacoustics/Spectrometry: Could optoacoustic spectrometry make noninvasive glucose monitoring viable?

A recent article in The Scientist that summarized the pursuit of noninvasive glucose monitoring quoted Vanderbilt University (Nashville, TN) physician and diabetes specialist Mark Rice as saying that the Holy Grail technology has not yet been invented, or at least not applied to glucose measurements.¹

Could extended near-infrared (near-IR) optoacoustic spectrometry (OAS) be the “it” to which Rice refers? The work of researchers from the German Research Centre for Environmental Health (Neuherberg, Germany) and the Technical University of Munich (Munich, Germany) indicates that it could (see figure).² Critically, this method—which leverages two deep-penetration techniques—overcomes a limitation that Rice identified as key: signal-to-noise ratio (SNR).

Among approaches proposed for noninvasive in vivo glucose monitoring, those based on spectroscopy show particular promise. Optical spectroscopy (OS) encompasses a variety of methods—including ultraviolet (UV), visible (Vis), mid-infrared (mid-IR), Fourier-transform infrared (FTIR), fluorescence, and Raman spectroscopy—commonly used to detect and assess composition and physical properties. In purely optical methods, however, photon scattering compromises the accuracy of spectral discrimination and quantification.

In OAS, a variant that uses an acoustic detector instead of an optical sensor, molecules in the specimen absorb transient light energy and return ultrasound signals. Because sound waves are less prone to scattering than light waves, they enable better penetration—a fact that has proven useful for noninvasive imaging and now holds promise for quantitative glucose sensing in deep tissue.

OAS has benefitted a range of quantitative biology applications and has even been used for blood-glucose measurement in the mid-IR, which benefits from the fact that glucose absorption identifies distinct signatures between about 800 and 1200 cm-1. But strong water absorption hampers the ability of mid-IR to analyze large aqueous volumes (including biological samples), and limits penetration to <100 μm.

Precise, sensitive, accurate

The German team applied and compared two methods for extracted glucose spectra—ratiometric and dictionary learning—with a training set of data and then validated using their probe in aqueous media across a wide range of glucose concentrations. While the dictionary learning method yielded an approximately 3.9-fold greater SNR with better correlation to predefined unit spectra, it requires careful calibration to maintain accuracy. The ratiometric approach produced somewhat lower SNR and correlation coefficient measurements, but gains speed by using data generated at two wavelengths, while also lowering cost and minimizing errors. The team noted that because SNR—which was above 11.4 across the board—relates to energy fluctuations of a tunable laser, they expect to further increase it by compensating for laser variability and using more stable optical sources.

The researchers’ findings explicitly showed a linear dependence between OA signal intensity and glucose concentrations (0–170 mg/dl), confirming that optoacoustic measurements attain sufficient sensitivity in the extended near-IR region. And, OAS’s highly sensitive and accurate readings generally agree with conventional OS, they said. They also investigated various spectral processing methods and examined which wavelength ranges perform best for glucose measurement within the extended near-IR. Thus, they demonstrated the feasibility of using near-IR optoacoustic spectroscopy to sense glucose with ±10 mg/dl precision.

The researchers’ OA spectroscope provides a framework for developing tools that can detect and accurately quantify glucose ex vivo and potentially translate that technology for in vivo application.

REFERENCES

1. C. Offord, The Scientist, article 50631 (2017).

2. A. Ghazaryan et al., Front. Endocrinol. (2018); https://doi.org/10.3389/fendo.2018.00112.