Prof. Shwetadwip Chowdhury (Department of Electrical and Computer Engineering, The University of Texas at Austin)

"Computational microscopy with multiple-scattering samples"

Abstract: Optical imaging is a major research tool in the basic sciences, and provides both morphological and molecular-specific imaging capabilities. Furthermore, it is the only imaging modality that routinely enables non-ionized imaging with subcellular spatial resolutions and high imaging speeds. In biological imaging applications, however, classical optical imaging is limited by tissue scattering to only a few hundred microns of imaging depth. This prevents visualization of most biological systems at system-wide scales.

Interestingly, though scattered light appears chaotic and random, it is deterministically related to 3D sample structure and can travel centimeters in tissue. If scattering can be fully corrected, it would be possible to image an order-of-magnitude deeper into biological tissue than currently possible, enabling noninvasive 3D imaging of thick and intact biological samples. This imaging capability would have broad implications for several scientific fields and will be a crucial first step towards large-scale 3D imaging of key biological systems and processes.

I present some recent computational imaging developments that aim to decode optical scattering. I show results that demonstrate that 1) optical scattering can be used to reconstruct a sample’s 3D phase and fluorescence at super-resolution levels; and 2) computational frameworks can unscramble scattered information and reconstruct 3D refractive-index of optically-thick samples. These developments utilize of large-scale computational nonlinear and nonconvex frameworks as well as novel computational microscope system designs.