Signal processing system to extract serum bilirubin concentration from diffuse reflectance spectrum of human skin

Abstract

Neonatal jaundice is a medical condition which occurs in newborns as a result of an imbalance between the production and elimination of bilirubin. Excess bilirubin in the blood stream diffuses into the surrounding tissue leading to a yellowing of the skin. The extra-vascular bilirubin act as interfering signal that limits the estimation of serum bilirubin from reflectance spectrum of human skin. This is particularly an issue for neonates who are being subjected to phototherapy (a common treatment for neonatal jaundice). Unfortunately, analytical models developed to study the light transport in human skin do not consider the effects of extra-vascular bilirubin (and other absorbing chromophores). A biomedical signal processing method that estimates serum bilirubin in presence of confounding signals such as melanin and extra-vascular bilirubin concentrations is presented. The new system model and nonlinear solver have been successful in estimating the serum bilirubin concentration on simulated spectral databases within an average error of 15%.