Our lab develops computational imaging methods, which jointly design optics, devices, signal processing and algorithms, and enable novel capabilities that each one alone cannot. Our research is inherently interdisciplinary, combining expertise in optical engineering, physics and computation. We work on imaging technologies for scientific, biomedical, and neuroscience applications.

Computational Microscopy for Neural Imaging 

Computational Miniature Mesoscope (CM2)

We develop “wearable” computational fluorescence microscope that achieves cm-scale FOV and µm-scale resolution with single-shot 3D imaging capability. A complete publication list is here.

Representative paper:


Deep learning for in-vivo neural signal modeling & extraction

We develop various deep learning techniques to enhance and analyze in-vivo neural signals. A complete publication list is here.

Representative paper:
Platisa, et al, High-speed low-light in vivo two-photon voltage imaging of large neuronal populations, Nature Methods 20pages1095–1103 (2023)



Computational Label-free Imaging

We work in the following major directions. A complete publication list is here.

Intensity Diffraction Tomography

We are developing high-speed computational 3D phase microscopy techniques by leveraging simple optical setups and advanced algorithms.

Representative paper:
Li, et al, High-speed in vitro intensity diffraction tomography, Advanced Photonics, 1(6), 066004 (2019).



Fourier Ptychography 

We are working computational microscopy techniques to achieve gigapixel phase imaging.

Representative paper:
Tian, et al, Computational illumination for high-speed in vitro Fourier ptychographic microscopy, Optica 2(10), 904-911 (2015).



Computational label-free chemical microscopy

We are working computational label-free microscopy techniques with chemically specific information.

Representative paper:
Zhao, et al, Bond-Selective Intensity Diffraction Tomography, Nat Commun 13, 7767 (2022).



Deep learning for Quantitative Bio-imaging

We work in the following major directions.  A complete publication list is here.

Cross-modality information transfer

We work on deep learning technique that allows knowledge transfer across different bio-imaging modalities.

Representative paper:
Cheng, et al, Single-cell cytometry via multiplexed fluorescence prediction by label-free reflectance microscopy, Science Advances  15 Jan 2021: Vol. 7, no. 3, eabe0431.




Reliable deep learning for quantitative bio-imaging

We develop uncertainty quantification techniques to provide more reliable deep learning predictions for quantitative bio-imaging.

Representative paper:
Xue, et al, Reliable deep learning-based phase imaging with uncertainty quantification, Optica 6, 618-629 (2019).



    Computational Metrology

    We developed computational imaging techniques for semiconductor metrology and inspection applications.

    Fourier Ptychograhpic Topography

    We develop novel topography techniques based on the reflection-mode Fourier ptychographic microscopy, termed Fourier ptychograhpic topography (FPT). FPT provides both a wide FOV and high resolution, and achieves nanoscale height reconstruction accuracy.

    Representative paper:
    H. Wang, et. al. Fourier ptychographic topography, Optics Express 31, 11007-11018 (2023).





    Computational Imaging in Complex Media

    We work in the following major directions.  A complete publication list is here.

    Multiple-scattering modeling & Physics-based deep learning

    We develop efficient and accurate multiple-scattering models to enable large-scale simulation of multiple-scattering in biological samples. These multiple scattering models form the foundation to develop physics-based deep learning models to achieve accurate 3D phase reconstructions. A complete publication list is here.

    Representative paper:
    Matlock, Zhu, Tian, Multiple-scattering simulator-trained neural network for intensity diffraction tomography, Optics Express 31, 4094-4107 (2023).



    Adaptive deep learning for descattering

    We work on adaptive deep learning framework to achieve robust descattering across a variety of scattering conditions.

    Representative paper:
    Tahir, et al, Adaptive 3D descattering with a dynamic synthesis network, Light: Science & Applications 11, 42, 2022.


    Deep Speckle Correlation

    We investigate “hidden information” in speckle information for imaging through scattering media.

    Representative paper:
    Li, Xue, Tian, Deep speckle correlation: a deep learning approach towards scalable imaging through scattering media, Optica 5, 1181-1190 (2018).


    Computational imaging with non-conventional optics

    Computational Imaging with Metasurface Photodetectors

    We develop computational imaging techniques for metasurface photodetectors to achieve non-conventional imaging capabilities. A complete publication list is here.

    Representative paper:
    Kogos, et al, Plasmonic ommatidia for lensless compound-eye vision, Nat. Communications 11: 1637 (2020).


    Computational diffuser funduscopy

    We develop a diffuser-based computational funduscope to enable low-cost, multi-functional, high-quality ocular imaging.

    A complete publication list is here.

    Representative paper:
    Li, et al, Diffuser-based computational imaging funduscope, Optics Express 28, pp. 19641-19654 (2020).