Optimal wavelength selection for optical spectroscopy of hemoglobin and water within a simulated light-scattering tissue

An algorithm that selects optimal wavelengths for spectral fitting of diffuse light reflectance spectra using a nonnegative least squares method is presented. Oxyhemoglobin, deoxyhemoglobin, and water are considered representative absorbers, but the approach is not constrained or limited by absorber selection provided native basis spectra are available. The method removes wavelengths iteratively from a scattering-modulated absorption matrix by maximizing the product of its singular values and offers considerable improvements over previously published wavelength selection schemes. Resulting wavelength selections are valid for a broad range of optical properties and yield lower RMS errors than other wavelength combinations. The method is easily modified and broadly applicable to tissue optical spectroscopy. Adaptation of the algorithm to select optimal light-emitting diodes for fitting blood is described.