Wavelength selection for imaging hemoglobin in skin

Hyperspectral imaging has been shown to be useful for producing qualitative maps of the concentration of hemoglobin in skin, and quantitative maps of its oxygen saturation. These maps provide information about the health of the skin, which may be useful in medical diagnosis and in planning intervention for skin lesions. Wavelengths have usually been determined on a somewhat intuitive basis using spectra based on analytical models. In the present work, we demonstrate a means of selecting wavelengths from a data cube of experimental data.