Strong reflection and scattering effects, arising at the boundaries of acoustically mismatched areas in living organisms, such as bones, lungs, and other air cavities, may introduce severe image artifacts into optoacoustic reconstructions. Yet, in many cases, an a priori knowledge on the location of strongly mismatched areas is available, either based on general anatomical knowledge or using other imaging modalities. In this letter, we suggest a statistical optoacoustic image reconstruction method, which uses a priori knowledge on the location of acoustic distortions in order to improve image quality and quantification. Significant improvements are showcased experimentally on tissue mimicking phantoms of different complexities.