Imaging Intelligence LaboratorySeoul National University
Imaging Intelligence LaboratorySeoul National University

seungahlee@snu.ac.kr
Department of Mechanical Engineering, Seoul National University

Research

Four areas of work: computational microscopy, including virtual staining of tissue slides; lensless cameras and learned optics; dynamic speckle imaging of sample dynamics such as blood flow; and interactive biology. Each section lists its projects, with papers, code and data on the theme pages.

  1. Computational Microscopy
  2. Lensless Cameras and Learned Optics
  3. Dynamic Speckle Imaging
  4. Interactive Biology

Computational Microscopy

2009–

Fourier ptychography (FP) illuminates a sample from many angles and combines the low-resolution images computationally into one wide-field, high-resolution image with quantitative phase. We have built smartphone, single-shot and polarization-resolved FP microscopes and use them for virtual staining of tissue slides.

Projects 7

Computational Microscopy: papers, code and projects

Pupil-plane polarization multiplexing FP microscope: LED matrix, sample, objective, split polarizer and camera.
Pupil-plane polarization multiplexing FP microscope: LED matrix, sample, objective, split polarizer and camera. H. Chae et al., Laser Photonics Rev. 2026

  • The same unstained cells in a raw low-resolution bright-field image (left) and in the FP quantitative phase image (right). IIL (conference presentation)

  • Quantitative phase of an unstained tissue section rendered as a 3D surface, where height shows optical thickness. IIL (conference presentation)

Virtual staining

2021–

We image unstained tissue slides with Fourier ptychographic microscopes and train neural networks to turn the measured phase into the appearance of chemically stained slides. Work with pathologists at Severance Hospital covers faded-slide restoration, faster scanning and stains other than H&E.

Projects 3

Virtual staining: papers, code and projects

Quantitative phase image of an unstained tissue region with glands.
Virtual H&E of the same region generated from the phase image.
Quantitative phase image of an unstained tissue region with glands. Virtual H&E of the same region generated from the phase image. Drag the handle to compare. IIL (conference presentation)

  • Whole-slide virtual H&E generated from a label-free FP phase image by unsupervised edge-guided virtual staining. IIL (conference presentation)

  • Virtual immunohistochemistry from label-free phase compared with chemical IHC. IIL (conference presentation)

Lensless Cameras and Learned Optics

2019–

A lensless camera replaces the lens with a thin phase mask above the image sensor and recovers the image computationally. We design the camera hardware, including the masks we fabricate, for each application, and develop the reconstruction algorithms. Applications include polarization, hyperspectral, eye-tracking and privacy-preserving imaging.

Projects 9

Lensless Cameras and Learned Optics: papers, code and projects

Real scenes imaged with the integrated lensless camera: reconstructed images (top), raw sensor images and the scenes (bottom).
Real scenes imaged with the integrated lensless camera: reconstructed images (top), raw sensor images and the scenes (bottom). Y. Lee, H. Chae et al., IEEE Photonics J. 2022

  • Fabricated phase mask held by tweezers, the lensless camera module next to a coin, and a side view showing its thickness. K. C. Lee et al., Optica 2023

Dynamic Speckle Imaging

2022–

We analyze speckle produced by dynamic scattering to measure the dynamics of a sample. Rolling-shutter speckle imaging measures blood flow from a single camera frame by using the row-by-row readout of a standard CMOS sensor. Applications include cerebral blood-flow mapping, pulse waveforms at the fingertip and real-time analysis with neural networks.

Projects 6

Dynamic Speckle Imaging: papers, code and projects

Video from one raw rolling-shutter speckle frame of a mouse cortex: a wipe reveals the blood-flow map computed from that single frame by row-by-row intensity correlation, then the view zooms into the elongated speckle grains. C. Yi et al., Optica 2022 (raw data, IIL-SNU/RSSI repository); map computed for this site

  • Mouse cortex maps from pixel-wise fitting, averaged fitting, TV-regularized fitting and a physics-constrained network. S. Byun et al., SPIE Photonics West 2025 (conference presentation)

  • Speckle-based eye-movement tracking: a rolling-shutter sensor with an elliptical aperture views laser speckle on the sclera; slanted speckles encode eye velocity. S. Byun et al., SPIE ABC 2025 (conference presentation)

Interactive Biology

2015–2022

Systems that couple a microscope and a projector so that people can interact in real time with living, light-responsive microorganisms, used in museum exhibits, installations, games and artworks.

Projects 5

Interactive Biology: papers, code and projects

MicroAquarium: body movements in front of the installation become light patterns inside the microscope, and the Euglena respond in real time. K. Lee, J. Jung, S. A. Lee, CHI EA 2020