Near-infrared Raman instrument for rapid and quantitative measurements of clinically important analytes

Abstract

We present the use of a near-infrared (IR) laser Raman spectroscopy instrument to measure the concentrations of many important analytes at their clinically relevant levels in the simulated human serum. The Raman signal is generated by a 745 nm diode laser in a disposable waveguide capillary cell that contains a submicroliter sample. The Raman spectrum is acquired from the sample in 10 s. The major error in quantitative Raman spectroscopy caused by the variation in laser power, optical alignment, and capillary cell size from measurement to measurement is eliminated by normalizing the spectrum to the dominant water peak at 3350 cm−1. Concentrations of glucose, acetaminophen, albumin, and other analytes are predicted using partial least squares (PLS) calibration. An effective multiple bandpass-filtering method was developed to enhance the signal of the desired analytes to interfering background ratio for improvement of PLS calibration accuracy. It is demonstrated that the accuracy of predicted concentrations for all analytes in the simulated human serum samples are highly acceptable for clinical diagnosis. The results promise the potential applications of the near-IR Raman instrument in medical practice.