Computational Imaging in Biology

Abstract

This dissertation presents an exploration of advanced computational imaging techniques aimedat enhancing biological imaging performance. Biological specimens often present challenges due to their sub diffraction limited complex structures and the scattering of light, which can obscure critical details. To address these issues, this work investigates the application of several novel computational techniques which enable sub – diffraction limited imaging, suppression of multiple scattering and correction of heavy optical aberrations, and label free contrast enhancement. Single pixel imaging allows for imaging at high speeds and extended wavelengths compared to traditional cameras. Pared together with computational and inverse problem techniques super resolution information can be extracted. Techniques for extracting higher resolution information from non-linear fluorescence induced temporal harmonics as well as fusion of quantum and classical correlations in photon arrival times are discussed. Tomographic reconstructions, iterative aberration correction via joint estimation and dual illumination single pixel reconstruction algorithms are also examined. Reflection matrix based imaging methods allow for diffraction limited imaging through deep tissue which induces large aberrations as well as multiple scattered photon background. Several reflection matrix based imaging techniques are applied as well as developed for use with SHG and THG holography which enables excellent label free contrast mechanism. Pairing SHG/THG holography with reflection matrix imaging allows for sub diffraction limited imaging with large contrast within deep tissue. Numerical simulations of multiple scattering were developed for better understanding of reflection matrix based imaging techniques as well as providing a test bed for new implementations. A multiple scattering model for efficient computation through large volumes of tissue was developed. using a multi-slab Rytov model. One such method which is tested numerically is transmission matrix “leap frogging”, as an attempt to reconstruct transmission matrix through tissue thicknesses which would otherwise be too thick for traditional measurement techniques. By integrating these methodologies, this research advances the state of biological imaging, enabling clearer, more detailed observations of cellular and subcellular structures. The results demonstrate the potential of these combined techniques to provide unprecedented insights into biological processes, thereby opening new avenues for research in fields such as cell biology, pathology, and biomedical engineering. Overall, this dissertation lays the groundwork for future innovations in computational imaging, ultimately contributing to a deeper understanding of complex biological systems.