Premium
Measurement of aqueous glucose in a model anterior chamber using Raman spectroscopy
Author(s) -
Lambert James L.,
Morookian John Michael,
Sirk Shan J.,
Borchert Mark S.
Publication year - 2002
Publication title -
journal of raman spectroscopy
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.748
H-Index - 110
eISSN - 1097-4555
pISSN - 0377-0486
DOI - 10.1002/jrs.871
Subject(s) - raman spectroscopy , aqueous solution , microscope , chemistry , analytical chemistry (journal) , confocal , raman microscope , partial least squares regression , analyte , chromatography , optics , raman scattering , physics , statistics , mathematics
Abstract Many laboratories have been developing techniques that might be applied toward a non‐invasive device that could optically measure aqueous humor glucose in the eye as a surrogate for blood glucose. We have configured a confocal Raman microscope that is capable of acquiring the Raman spectra of the aqueous humor in vivo . However, because of the potential for optical toxicity, a major design consideration is evaluating the tradeoff between the energy of exposure and measurement accuracy. Toward this end, we have developed a physical model of the anterior chamber in the eye for studying the feasibility of using a confocal Raman microscope for determining the concentration of glucose and other metabolites. The Raman spectrum of aqueous humor closely resembles the spectrum of its four primary Raman scatterers, (glucose, urea, lactate and ascorbate) mixed in normal saline. Aqueous mixtures of these analytes were placed under a contact lens resting on a quartz plate to form a physical model of the anterior chamber filled with aqueous humor. Calibration standards were prepared by non‐collinearly varying the measured concentrations of the analytes over a range of 0–13 times the mean physiological levels found in aqueous humor. A confocal microscope was used to acquire spectra using an excitation wavelength of 785 nm. The accuracy to which the concentration with analytes could be optically measured was evaluated using the partial least‐squares (PLS) algorithm and spectral datasets collected using exposure energies of 75, 150, 300, 900, and 1800 mJ. The optimum standard error of prediction (SEP) obtained for glucose was 34.3 mg dl −1 (2.72% of full range) using a 12‐factor PLS model calculated using intensity‐normalized spectra acquired over a free spectral range of 400–1800 cm −1 with 900 mJ excitation energy. Clinically acceptable predictability (4.6%) was obtained with 300 mJ total excitation energy. Similar results were obtained for urea (SEP = 8.84 mg dl −1 , 1.88%), ascorbate (SEP = 17.1 mg dl −1 , 8.22%) and lactate (SEP = 82.9 mg dl −1 , 7.59%) using 900 mJ exposure. Our results suggest that Raman spectroscopy may provide a feasible method of non‐invasive glucose measurement if ocular toxicity can be avoided. Copyright © 2002 John Wiley & Sons, Ltd.