Related Experiment Video
Updated: Jan 9, 2026
09:31
High-Speed Ultraviolet Photoacoustic Microscopy for Histological Imaging with Virtual-Staining assisted by Deep Learning
Published on: April 28, 2022
3.5K
VS-FPM: Large-Format, Label-Free Virtual Histopathology Microscopy.
Christopher Bendkowski1, Adam P Levine2,3, Manuel Rodriguez-Justo2,3
1UCL Hawkes Institute and Department of Computer Science, University College London, London, UK.
BME Frontiers
|December 4, 2025
Summary
Virtual staining using Fourier ptychographic microscopy (VS-FPM) creates realistic H&E images from unstained tissues. This label-free digital pathology method enables accurate diagnoses and overcomes limitations of traditional histopathology.
Area of Science:
- Digital Pathology
- Microscopy
- Machine Learning
Background:
- Virtual staining (VS) generates realistic histological images from label-free data.
- VS can streamline workflows, improve consistency, and enable novel tissue analysis.
- Fourier ptychographic microscopy (FPM) offers high resolution and large fields of view.
Purpose of the Study:
- To develop and assess a novel VS method (VS-FPM) using supervised machine learning.
- To generate brightfield H&E images from FPM phase images of unstained tissues.
- To evaluate VS-FPM for colonic polyp diagnosis.
Main Methods:
- Trained a conditional generative adversarial network to translate FPM images to H&E images.
- Utilized unstained tissue samples for image acquisition.
- Assessed diagnostic accuracy using colonic polyp cases.
Main Results:
- VS-FPM achieved spatial resolution comparable to traditional scanners.
- VS-FPM images closely resembled chemically stained H&E images.
- Pathologists accurately diagnosed normal vs. dysplastic tissues using VS-FPM.
Conclusions:
- VS-FPM is a reliable and accessible virtual staining method.
- The technique overcomes limitations of conventional histopathology microscopy.
- VS-FPM offers advantages for label-free digital pathology.
