ECE PhD Thesis Defense: Bingying Zhao

Title: Techniques for High-speed High-resolution in-vivo Microscopic Imaging

Presenter: Bingying Zhao

Advisor: Professor Jerome Mertz

Chair: Professor Anna Swan

Committee: Professor Jerome Mertz, Professor Lei Tian, Professor Tianyu Wang, Professor Darren Roblyer

Google Scholar Link: ‪ https://scholar.google.com/citations?hl=en&user=UiBAqeUAAAAJ

Abstract: Optical microscopy has long been a cornerstone for exploring biological and biomedical systems. 3D microscopy with high spatialtemporal resolution over large fields of view is increasingly essential in life sciences. This thesis proposes two innovations to address this need: (1) high-resolution multi-Z confocal microscopy using a diffractive optical element (DOE), (2) an image deblurring algorithm termed Deblurring by Pixel Reassignment (DPR). Together, these methods aim to expand the capabilities of optical microscopy for high-speed, high-resolution, and quantitative imaging across complex biological samples.

In the first part, we present a high-resolution multi-Z confocal microscopy system that overcomes the resolution limitations of earlier designs. By introducing a DOE to generate multiple high-NA excitation foci, axially aligned with corresponding reflective detection pinholes, the system achieves simultaneous multi-plane imaging without compromising spatial resolution. This microscope design significantly enhances both lateral and axial resolution compared to previous low-NA implementations, while preserving optical sectioning and maximizing photon collection efficiency. We demonstrate the performance of this microscope through high-speed, high-resolution imaging of dynamic biological processes across multiple systems.

In the second part, we introduce DPR, a computational algorithm designed to enhance the spatial resolution of fluorescence microscopy through a simple yet effective pixel reassignment strategy. Instead of relying on complex deconvolution models, DPR operates under minimal assumptions, using only local gradient information and the premise that the emission point spread function (PSF) is centered at its peak. By reassigning pixel intensities along the gradient direction, DPR effectively sharpens images, improves the separation of closely spaced fluorophores, and enhances structural clarity even beyond the conventional diffraction limit. Its simplicity, versatility across imaging conditions, and robustness to noise make DPR a powerful post-processing tool for advancing super-resolution imaging in both dense and dynamic biological samples.

Finally, we extend DPR to multi-dimensional fluorescence imaging, focusing on fluorescence lifetime imaging microscopy (FLIM). By applying the DPR frame-by-frame to the lifetime series, we achieve spatial sharpening, enabling the creation of superresolved fluorescence lifetime maps.