Two photoacoustic spectral coloring compensation techniques adapted to the context of human <i>in-vivo</i> oxygenation measurements

Photoacoustic imaging can potentially map oxygen saturation (sO 2 ) non-invasively. However, in-vivo human application is challenging due to spectral coloring, which causes a wavelength-dependent fluence attenuation and uncertainty in the estimation of chromophore concentrations deep in tissue. This study compares the performances of two previously proposed methods for spectral coloring compensation on in-vivo human data. Both methods have been modified and adapted to this context. The first modified method was evaluated using a tissue-mimicking phantom, showing restoration of the original spectrum of the target and decreasing the relative mean square error from 65% to 1.2% for the highest concentration. Spatial maps of sO 2 were estimated from in-vivo human finger measurements using both methods and compared with linear unmixing. Both methods reconstructed comparable values of sO 2 and reduced depth-dependent changes in sO 2 , typically seen with linear unmixing, resulting in a gradient of saturation closer to zero as expected physiologically.