Advancing intensity diffraction tomography with multiple scattering models in transmission and reflection systems
Date
2025
DOI
Authors
Version
OA Version
Citation
Abstract
This dissertation advances intensity diffraction tomography (IDT), a computational imaging technique that enables label-free, non-invasive 3D imaging. IDT directly measures scattered light intensity using a standard microscope equipped with a programmable LED array. Without requiring interferometry or mechanical scanning, it provides a stable and economical approach for high-resolution 3D quantitative phase imaging (QPI). Recent advances in IDT have demonstrated promising capabilities beyond basic phase imaging, including high-speed acquisition for dynamic imaging, enhanced resolution through darkfield illumination, and chemical specificity via bond-selective IDT (BS-IDT), making it increasingly valuable for biomedical applications. However, while hardware innovations continue to emerge, traditional IDT reconstruction algorithms based on linear single-scattering models are limited to simple, weakly scattering samples. This dissertation advances IDT by developing robust computational methods capable of handling complex multiple-scattering samples while maintaining the technique's practical advantages. After reviewing recent hardware developments and applications, the first part of this work focuses on developing the efficient split-step non-paraxial (SSNP) multiple-scattering model that accurately describes light-sample interactions across diverse scenarios, particularly for thick and strongly scattering samples under high-NA illumination. I present both model-based iterative optimization and deep learning approaches for solving the associated large-scale inverse scattering problem, demonstrating improved reconstruction quality and computational efficiency for various biomedical samples. The second part introduces reflection-mode IDT for imaging samples on reflective substrates, expanding the technique's applications to semiconductor inspection and industrial metrology. This includes adapting the multiple-scattering model for reflective geometries, developing specialized reconstruction algorithms for single- and multi-layer samples, and introducing novel illumination angle calibration methods tailored for reflection setups. The developed reflection-mode system demonstrates high-fidelity volumetric imaging capabilities valuable for industrial inspection, successfully resolving complex multi-layered structures with sub-micron resolution across millimeter-scale fields of view. These advances significantly expand IDT's capabilities for 3D QPI of complex samples while preserving its fundamental advantages of simplicity, stability, and cost-effectiveness. As the field continues to evolve, opportunities remain to further improve reconstruction fidelity, integrate new imaging modalities, and expand applications across both biomedical research and industrial inspection.
Description
2025
License
Attribution-ShareAlike 4.0 International