Improved Myoglobin Saturation Measurement Made by Partial Least-Squares Analysis of Optical Reflectance Spectra

Abstract

Accurate myoglobin oxygen saturation determination from tissue using optical spectral analysis has been limited by overlap in spectral features between myoglobin and hemoglobin. Partial least-squares analysis of reflectance spectra has been successfully used to predict myoglobin oxygen saturation from blood perfused hearts, although uncertainty remains as to how much error in the myoglobin saturation estimate occurs due to hemoglobin contamination. This study was undertaken to quantify the error due to hemoglobin in the partial least-squares estimate of myoglobin saturation. Calibration spectra were developed by mathematical addition of individually acquired in vitro reflectance spectra from various absorbing species in scattering media. The isolated perfused guinea pig heart model was used to allow for switching between crystalloid and red blood cell perfusion in a beating heart. Inclusion of cytochrome c and cytochrome oxidase in a calibration set containing hemoglobin and myoglobin improved the accuracy of the partial least-squares analysis of myoglobin saturation. Both the Mahalanobis distance test and the residual ratio test demonstrated improved representation of spectra obtained from the heart using a calibration set that included both cytochromes. A 4.3% error in myoglobin oxygen saturation measurements was determined.