Abstract
Light-sheet fluorescence microscopy (LSFM) enables fast, light-efficient 3D imaging of large, optically cleared specimens, but its image quality is limited by stripe artifacts and anisotropic, diffraction-limited resolution, hindering quantitative investigation of cellular-level biology. This project aims to develop a computational framework for LSFM to achieve isotropic super-resolution in 3D. First, a forward operator will be designed based on simulation of scattered illumination patterns and spatially-variant detection point spread functions. Second, a hybrid iterative super-resolution solver combining physics-based modelling with deep-learning-based inverse optimization based on multi-angle LSFM data will be developed to balance fidelity and enhancement. Third, temporal and spectral variations in fluorescence emission will be integrated to further improve resolution. The framework will be benchmarked using simulations and real phantoms, and will be tested on optically-cleared Huntington's disease (HD) mouse brains – as an important application that truly demands comprehensive mapping of pathological defects with cellular precision. Our approach will shift LSFM-based organ-level research from coarse regional measures to cell-resolved quantitative analysis, while remaining broadly applicable across LSFM platforms and biological studies.
Researcher(s)
Research team(s)
Funding
Project type(s)