Image reconstruction in cross-sectional optoacoustic tomography based on non-negative constrained model-based inversion

In optoacoustic tomography, images representing the light absorption distribution are reconstructed from the measured acoustic pressure waves at several locations around the imaged sample. Most reconstruction algorithms typically yield negative absorption values due to modelling inaccuracies and imperfect measurement conditions. Those negative optical absorption values have no physical meaning and their presence hinders image quantification and interpretation of biological information. We investigate herein the performance of optimization methods that impose non-negative constraints in model-based optoacoustic inversion. Specifically, we analyze the effects of the non-negative restrictions on image quality and accuracy as compared to the unconstrained approach. An efficient algorithm based on the projected quasi-Newton scheme and the limitedmemory Broyden-Fletcher-Goldfarb-Shannon method is used for the non-negative constrained inversion. We showcase that imposing non-negative constraints in model-based reconstruction leads to a quality increase in cross-sectional tomographic optoacoustic images.