Author ORCID Identifier
0009-0000-0891-8607
Document Type
Dissertation
Date of Award
5-31-2024
Degree Name
Doctor of Philosophy in Computer Engineering - (Ph.D.)
Department
Electrical and Computer Engineering
First Advisor
Xuan Liu
Second Advisor
Kevin D. Belfield
Third Advisor
Leonid Tsybeskov
Fourth Advisor
Dong Kyun Ko
Fifth Advisor
Yuanwei Zhang
Abstract
Microscopy plays a crucial role across various scientific fields by enabling structural and functional imaging with microscopic resolution. In biomedicine, microscopy contributes to basic research and clinical diagnosis. Conventionally, optical microscopy derives its contrast from the amplitude of the optical wave and provides visualization of the physical structure of the sample qualitatively. To understand the function at the cellular or tissue level, there is a need to characterize the sample quantitatively and explore contrast mechanisms other than light intensity. Image enhancement or reconstruction from microscopic imaging systems is known as computational microscopy, and it involves the application of computational techniques and algorithms. By integrating concepts from signal processing, computer vision, and optics, it overcomes the constraints of conventional microscopy methods to get comprehensive information from unprocessed image data. Improved image quality, higher resolution, and the ability to use new imaging modalities are all made possible by computational microscopy. Notable forms of computational microscopy include super-resolution fluorescence microscopy, quantitative phase imaging, and Fourier ptychography. Computational microscopy is at the intersection of computer science and optics. This dissertation summarizes my research in functional phase microscopy based on optical computation, as well as deep learning assisted image analysis for quantitative phase imaging and optical coherence tomography (OCT) imaging.
Phase microscopy utilizes the phase of coherent light to extract significant information from the sample that is not available from amplitude imaging. Phase-resolved imaging, particularly quantitative phase microscopy, has advanced significantly in recent years, due to the development in the light source and light detector. This dissertation develops a full-field Doppler phase microscopy (FF-DPM) technology based on an innovative optical computation strategy that enables depth-resolved imaging and phase quantification. FF-DPM can image subtle mechanical motion at cellular and sub-cellular scales and can be used to study how extracellular particles interact with cultured cells and more generally how cells interact with their environment.
With technological advancement, current imaging systems can achieve a high imaging speed which is critical for comprehensive sample characterization and functional imaging of dynamic processes. However, the capability to perform high-speed imaging also implies a massive volume of data acquired. It is challenging to process the data and extract meaningful information from the data. Here this dissertation investigate deep learning approaches to obtain high-level information from the raw data. After acquiring the image, it is also important to analyze and process the image. An artificial intelligence assistant image analysis method is developed for Optical coherence tomography. Optical coherence tomography (OCT) is a non-invasive imaging technique used for visualizing cross-sectional views of tissue. Previous studies have demonstrated that OCT could reveal morphological characteristics of breast tissues with different pathologies. Despite technological development in the past decade, it remains challenging for a human reader to perform high-level tissue characterization using OCT data in clinical tasks, such as the differentiation of tumors from normal breast tissue. A dual-modality OCT characterization (volumetric OCT imaging and quantitative optical coherence elastography) on human breast tissue specimens was performed. Then, a U-Net for automatic image segmentation was trained and validated .
By addressing these challenges and offering insights, this dissertation contributes to the function of phase microscopy and OCT data analysis. In future research, the combination of convolutional neural network and full-field Doppler phase microscopy (FF-DPM) technology is promising. A system that can automatically reconstruct the profile of multiple cells should be investigated.
Recommended Citation
Liu, Yuwei, "Computational microscopy for biomedical imaging with deep learning assisted image analysis" (2024). Dissertations. 1758.
https://digitalcommons.njit.edu/dissertations/1758
Included in
Biomedical Engineering and Bioengineering Commons, Electrical and Electronics Commons, Optics Commons
