A number of variables influence the transcutaneous measurement of bilirubin in the blood by reflectance spectrophotometry. By comparing the reflectance spectra simulated by a Monte-Carlo model with the spectra measured on 78 bilirubinemic infants, we analyze the origin and magnitude of several important sources of error. The results indicate that variations in blood volume, hemoglobin oxygen saturation, and scattering are the dominate variables that contribute to instrumental errors, defined in this study as errors that result from the imperfect selectivity of the instrument to the absorption of bilirubin in a homogeneous tissue medium. We found that these errors can be reduced by the proper choice of wavelengths in two-and four-wavelength algorithms based on generalized first-and secon...[ Read more ]

A number of variables influence the transcutaneous measurement of bilirubin in the blood by reflectance spectrophotometry. By comparing the reflectance spectra simulated by a Monte-Carlo model with the spectra measured on 78 bilirubinemic infants, we analyze the origin and magnitude of several important sources of error. The results indicate that variations in blood volume, hemoglobin oxygen saturation, and scattering are the dominate variables that contribute to instrumental errors, defined in this study as errors that result from the imperfect selectivity of the instrument to the absorption of bilirubin in a homogeneous tissue medium. We found that these errors can be reduced by the proper choice of wavelengths in two-and four-wavelength algorithms based on generalized first-and second-derivative processing, respectively. Bilirubin measurements obtained from neonates in a clinical setting suggest that the low predicted RMS error for bilirubin concentration (less than plus or minus]6 μmol L^-l) does not appear to be achievable in practice with present methods, because the bilirubin in the tissue volume probed by the reflected light is not distributed uniformly at the same concentration as in the blood. Variations in the ratio of the concentrations of bilirubin in the tissue and blood contribute an additional error that, in large part, overwhelms the contribution of instrumental error. Actual measurement limits, with these effects included, were found to be approximately (less than plus or minus]20 μmol L^-l). We conclude that the benefits that can be gained through improvements in the measurement algorithm-although substantial-do not appear to be sufficient to reach the ultimate goal of accurate non-invasive measurement of bilirubin concentration in the blood. New methods need to be developed to measure absorption of bilirubin in the blood directly instead of in the bulk tissue.

Fractional derivative spectroscopy, optimal probe geometry for spectroscopy of biological tissue and new algorithm for bilirubinometry were also discussed in the study.