Exploring the nonconvex landscape of biological inverse-scattering

Abstract

Optical imaging is a major research tool in the basic sciences, but is limited by tissue scattering to short imaging depths. This prevents large-scale bio-imaging by allowing visualization of only the outer superficial layers of an organism, or specific components isolated from within the organism and prepared in-vitro. Our lab develops inverse-scattering techniques to computationally correct for the effects of tissue scattering and reconstruct high-resolution volumetric images of optically-thick scattering sample. However, these methods are nonconvex and susceptible to local minima. We demonstrate that reconstruction quality is critically, and sometimes counterintuitively, influenced by factors such as the choice between field- and amplitude-based cost functions, initialization strategies for the sample volume, and the application of refocusing. Lastly, we present the methodology that we found through empirical testing has proven most effective for robust and high-quality 3D imaging of diverse biological scattering samples.