| Author: | Cheng, Shengfu |
| Title: | Computationally assisted deep-tissue optical imaging and patterned light delivery |
| Advisors: | Lai, Puxiang (BME) |
| Degree: | Ph.D. |
| Year: | 2024 |
| Department: | Department of Biomedical Engineering |
| Pages: | xix, 120 pages : color illustrations |
| Language: | English |
| Abstract: | Optical imaging has its unique values in biomedicine, as light is non-radiative and highly sensitive when interacting with biological tissues, offering structural and functional information with high spatial resolution. The biological tissues are, however, optically turbid with heterogeneous refractive index, causing multiple light scattering that prevents high-resolution optical focusing and imaging at depths in tissue. Over the past decades, many optical technologies have been developed that can overcome, reduce, or bypass strong light scattering for deep- tissue imaging. For example, wavefront shaping has been developed to overcome strong optical scattering and focus light through complex media, making it possible for high-resolution imaging within or through thick tissue. Photoacoustic imaging transforms the strongly scattered light into weakly scattered sound for detection, thus allowing a larger penetration depth with tunable spatial resolution. Fiber-optic endoscopic imaging can bypass optical scattering for in-situ microscopic observation with minimal invasion, especially when wavefront shaping is implemented with a multimode fiber (MMF) or fiber bundle. With photoacoustic imaging and the light control of MMF as the two primary technologies here, this thesis presents a comprehensive study aimed at pushing the boundaries of deep-tissue optical imaging and promoting light delivery related applications in complex media. The research work of the thesis can be divided into three major parts. The first part (Chapter 2) is to explore deep-tissue optical-resolution photoacoustic microscopy (OR-PAM), which is based on image transformation on blurry yet “seeing-deeper” images of an acoustic-resolution microscopy (AR-PAM) through deep learning. The trained network successfully transformed AR-PAM images into high-resolution counterparts, with the lateral resolution improved from 54.0 µm to 5.1 µm. Further, it significantly improved a mouse deep-brain image and retained high imaging resolution at tissue depths beyond one optical transport mean free path, which marks a big leap towards deep-tissue OR-PAM. The second part (Chapters 3 and 4) delves into the control of light delivery through MMF for deep-tissue applications based on wavefront optimization or retrieval of transmission matrix (TM). To be specific, by adopting a natural gradient ascent-based wavefront shaping strategy, coupled with a novel fitness function based on cosine similarity, we achieved high-quality pattern projection through a 15-meter-long unstable MMF. Apart from focusing, we also propose an alternating projection-based phase optimization method for suppressing scattered light in an arbitrary target region with a fast frame rate. In addition, we also study TM reconstruction from output intensity measurement by developing a modified nonconvex optimization method, which demonstrates optimum efficiency with reduced running time or sampling ratio. The ability to retrieve a large-scale TM of MMF in a reference-less setup opens new avenues for deep-tissue optical imaging and manipulation, etc. In the last part (Chapter 5), we investigate MMF-based fluorescence imaging and especially explore the online calibration of MMF. We have confirmed the performance of point- scanning-imaging on the samples of microspheres and mouse brain slice via MMF calibration in the transmission mode. Besides that, a fluorescence -based intensity TM was proposed for reflection-mode calibration of MMF, which is feasible in principle but needs further experimental validation. These studies serve a common goal of advancing the field of deep-tissue optical imaging and focusing. Although the part of PA imaging seems to be less relevant to MMF-based focusing and imaging, the wavefront control methods we have developed may also be applied to enhance the focusing quality and AR-PAM imaging in deep tissues physically. In summary, the thesis presents several computational and optimization methods that address key challenges faced by deep-tissue optical imaging and focusing, such as the limited resolution of AR-PAM and precise light control through complex media etc. Our research is valuable to deep-tissue PA imaging and MMF-based endoscopy, promising light delivery-related biomedical applications such as deep-tissue phototherapy and optogenetics. Through the above works, we have demonstrated that the integration of computational approaches with optical imaging technologies can unlock new potentials towards seeing deeper with greater clarity into biological tissues and even living bodies. |
| Rights: | All rights reserved |
| Access: | open access |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/13237