Bayesian Cramér-Rao Bound Optimization of the Illumination Pattern in Quantitative Photoacoustic Computed Tomography

Quantitative photoacoustic computed tomography (PACT) is an emerging biomedical imaging modality that leverages the photoacoustic effect to reconstruct images of underlying molecular or optical properties of biological tissue without ionizing radiation. The technology has the potential to improve the diagnosis and treatment of physiological diseases such as breast cancer by enabling high-contrast imaging of tissue metabolism and angiogenesis. However, there are currently no established design standards for quantitative PACT imaging, and current systems often require the use of multiple illumination sources due to the ill-posed nature of the image reconstruction problem. This motivates the need for virtual imaging studies to explore and optimize the design of PACT imagers. This work introduces a Bayesian Cramér-Rao bound based optimal experimental design (OED) approach for quantitative PACT imaging that enables the computationally efficient ranking of different illumination patterns without the ethical concerns and expense of human subject studies. The approach was validated using a simulation study with a stylized two-dimensional imaging system. The results demonstrate that the Bayesian Cramér-Rao bound is a computationally efficient surrogate for the expected reconstruction error in optimizing quantitative PACT imaging systems.