Optimal wavelength combinations for near-infrared spectroscopic monitoring of changes in brain tissue hemoglobin and cytochrome c oxidase concentrations.

We analyze broadband near-infrared spectroscopic measurements obtained from newborn piglets subjected to hypoxia-ischemia and we aim to identify optimal wavelength combinations for monitoring cerebral tissue chromophores. We implement an optimization routine based on the genetic algorithm to perform a heuristic search for discrete wavelength combinations that can provide accurate concentration information when benchmarked against the gold standard of 121 wavelengths. The results indicate that it is possible to significantly reduce the number of measurement wavelengths used in conjunction with spectroscopic algorithms and still achieve a high performance in estimating changes in concentrations of oxyhemoglobin, deoxyhemoglobin, and oxidized cytochrome c oxidase. While the use of a 3-wavelength combination leads to mean recovery errors of up to 10%, these errors drop to less than 4% with 4 or 5 wavelengths and to even less than 2% with 8 wavelengths.