Computational high-throughput phase imaging

Hu, Chenfei

Permalink

https://hdl.handle.net/2142/112967 Copy

Description

Title

Computational high-throughput phase imaging

Author(s)

Hu, Chenfei

Issue Date

2021-06-22

Director of Research (if dissertation) or Advisor (if thesis)

Popescu, Gabriel

Doctoral Committee Chair(s)

Popescu, Gabriel

Committee Member(s)

• Gruev, Viktor
• Kong, Hyunjoon
• Lam, Fan

Department of Study

Electrical & Computer Eng

Discipline

Electrical & Computer Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• quantitative phase imaging
• computational imaging
• harmonic optical tomography
• phase imaging with computational specificity
• cell viability
• synthetic aperture
• space-bandwidth product

Abstract

Quantitative phase imaging (QPI) is a remarkable modality for its wide range of potentials in studying biological structures without exogenous labeling. QPI utilizes the optical phase delay across the sample, instead of the image intensity, as the image contrast mechanism to provide highly sensitive information on transparent specimens (e.g., cells and tissue). Even though tremendous progress in this field has been achieved in the past 1-2 decades, some challenges remain to be addressed. On one hand, due to the lack of chemical specificity, this label-free modality cannot identify particular structures in the sample of interest. In addition, the throughput of the modality is limited by the hardware configuration. This Ph.D. training is centered around developing advanced QPI platforms to address the aforementioned challenges. This dissertation provides a summary of the main research results. First, we studied theories of light-matter interaction and derived physical models of light scattering, which provide guiding principles to develop image platforms. To increase the chemical specificity, we extended the principles of 3D QPI to nonlinear optical processes and demonstrated harmonic optical tomography (HOT), a quantitative 3D imaging platform that measures χ(2) from biological materials. HOT reports the first theoretical and demonstration of solving the inverse problem in nonlinear media without scanning the objects. Besides, by employing artificial intelligence, we demonstrated instantaneous viability assessment of unlabeled cells using phase imaging with computation specificity (PICS). This new concept utilizes deep learning techniques to extract information within the subcellular structure of specimens measured by QPI. This new approach reports an unbiased assessment of cells with more than 95% accuracy, without concerns on toxicity associated with the chemical stains. To improve the throughput of conventional QPI microscopes, we also developed synthetic aperture interference light (SAIL) microscopy as a solution for high-SBP, label-free imaging of micro- and mesoscopic objects. This imaging system directly measures the optical pathlength map, which removes the need for computational phase retrieval algorithms used in conventional methods. SAIL reduces the complexity of the reconstruction and data redundancy, while the results can be computed by a CPU processor in a matter of seconds.

Graduation Semester

2021-08

Type of Resource

Thesis

Permalink

http://hdl.handle.net/2142/112967

Copyright and License Information

Copyright 2021 Chenfei Hu

Owning Collections