Massachusetts Institute of Technology Spectroscopy Home   search
Research in Biomedical Optics

Non-Invasive Measurement of Blood Analytes using Raman spectroscopy

Pioneering work at the MIT Spectroscopy Laboratory

Our laboratory has pioneered the use of Raman spectroscopy for biological applications [Buschman 2001, Haka 2002, Shafer-Peltier 2002, Motz 2004, Haka 2005, Motz 2005, Haka 2006, Motz 2006, Scepanovic 2006, and others], including measurement of blood analytes [Berger 1997, Berger 1998, Berger 1999, Koo 1999, Hanlon 2000, Enejder 2002, Enejder 2005]. Our initial demonstration of concentration measurements in biological media reported measurement of multiple analytes including glucose, urea, cholesterol, triglyceride, albumin, total protein and hematocrit in serum and whole blood samples from sixty-nine patients over a seven-week period. The serum measurement errors were within clinical accuracy requirements, but whole blood measurement errors were considerably higher.

Our data indicated that the reduced signal levels obtained from whole blood samples reported by Berger et al. [1999] were the principal source of increased error in the study. Accordingly, we improved the signal collection capability of our instrument by a factor of >4. A subsequent whole blood study using the updated instrument demonstrated the feasibility of measuring multiple analytes in whole blood with clinical accuracy [Enejder 2002]. In this study, whole blood samples were collected from 31 patients undergoing routine clinical evaluation. For each sample, 30 consecutive 10-second spectra were collected over a 5-minute period. Conventional clinical laboratory methods were used to measure eight reference analyte concentrations, including glucose, urea, total protein, albumin, triglycerides, hematocrit and hemoglobin. These reference concentrations were correlated with the recorded Raman spectra through multivariate calibration and validation. Table 1 of Enejder et al. [2005] summarizes the results of PLS cross validation for the whole blood data set. All analytes showed strong correlation between the predicted and the reference concentrations (r2 values >0.90), except for total cholesterol (r2 = 0.66).

Based on the promising results of the ex vivo whole blood study described above, we conducted an initial in vivo evaluation of the ability of Raman spectroscopy to measure glucose transcutaneously. In this study, a series of Raman spectra were collected from the forearms of healthy human volunteers in conjunction with an oral glucose tolerance test (OGTT). Raman spectra and reference blood glucose concentrations were measured frequently during the 2-3 hour duration of test. A Hemocue® glucose analyzer provided the reference measurement for the blood analysis. A calibration model was generated individually from the data from each volunteer using PLS with leave-one-out cross validation. For the collected data from all 17 volunteers, the mean absolute error was 7.8% and the r2 was 0.87.  Although single-subject cross validation resulted in acceptable error estimates, prospective prediction on multiple subjects proved more difficult. Given that differences in skin composition are a major experimental variable between different subjects, we hypothesized that the increase in prediction error between individuals is due to the substantial turbidity variations observed across different subjects in the study.

  1. Barman I, Kong CR, Singh GP, and Dasari RR, “Effect of photobleaching on calibration model development in biological Raman spectroscopy", J Biomed Opt 15: 10134SSR, (2011).
  2. Barman I, Kong CR, Dingari NC, Dasari RR, and Feld MS, “Development of robust calibration models using support vector machines for spectroscopic monitoring of blood glucose,” Anal. Chem. 82(23):9719-9726, (2010).
  3. Barman I, Kong CR, Singh GP, Dasari RR and Feld MS, “Accurate spectroscopic calibration for noninvasive glucose monitoring by modeling the physiological glucose dynamics,” Anal. Chem. 82(14):6104-6114, (2010).
  4. Barman I, Singh GP, Dasari RR and Feld MS, “Turbidity-corrected Raman spectroscopy for blood analyte detection,” Anal. Chem. 81(11):4233-4240, (2009).
  5. Barman I, Singh GP, Dasari RR, Feld MS. “A novel Raman spectral correction scheme for quantitative analysis of blood analytes in turbidity media”, Photonics West, San Jose, BIOS (2009).
  6. Shih WC, Bechtel KL and Feld MS. “Intrinsic Raman spectroscopy for quantitative biological spectroscopy Part I: Theory and simulations,” Opt. Express 16(17): 12726-12736, (2008).
  7. Bechtel KL, Shih WC and Feld MS. “Intrinsic Raman spectroscopy for quantitative biological spectroscopy Part II: Experimental applications,” Opt. Express 16(17): 12737-12745, (2008).
  8. Singh GP, Barman I, Kong CR, Dasari RR, Feld MS. “Application of shifted excitation Raman difference technique for quantitative biological Raman spectroscopy”, Gordon Research Conference: Vibrational Spectroscopy, South Hadley, (2008).
  9. Barman I, Singh GP, Dasari RR, Feld MS. “Transcutaneous measurement of blood analyte concentration using Raman spectroscopy”, International Conference on Perspectives in Vibrational Spectroscopy, Trivandrum, India, (2008)
  10. Shih WC, Bechtel KL and Feld MS. “Constrained Regularization: Hybrid Method for Multivariate Calibration,” Anal. Chem. 79 (1): 234-239, (2007).
  11. Enejder AMK, Scecina, TG, Oh, J, Hunter M, Shih WC, Sasic S, Horowitz G and Feld MS. "Raman Spectroscopy for non-invasive glucose measurements," J Biomed Opt 10: 031114, (2005).