Related Experiment Video
Updated: Jul 30, 2026
13:49
High-resolution Fiber-optic Microendoscopy for in situ Cellular Imaging
Published on: January 11, 2011
34.5K
Calibration-free quantitative phase imaging in multi-core fiber endoscopes using end-to-end deep learning
Optics Letters
|January 9, 2024
Summary
A new learning-based quantitative phase imaging (QPI) method enables real-time endoscopic imaging using multi-core fibers (MCFs). This breakthrough significantly speeds up phase reconstruction, allowing for video-rate imaging in challenging environments.
Area of Science:
- Biomedical Optics
- Medical Imaging
- Computational Imaging
Background:
- Quantitative phase imaging (QPI) offers label-free, in vivo endoscopic imaging.
- Conventional QPI methods face computational limitations for real-time applications.
- Multi-core fibers (MCFs) present a minimally invasive imaging pathway.
Purpose of the Study:
- To develop a learning-based QPI method for MCFs to overcome computational bottlenecks.
- To achieve video-rate imaging speeds for enhanced endoscopic visualization.
- To create an open-source dataset for training and validating MCF phase imaging algorithms.
Main Methods:
- Implemented a deep neural network (DNN) for rapid phase reconstruction.
- Developed an optical system for automated dataset generation.
- Utilized a dataset of 50,176 paired speckles and phase images for training.
Main Results:
- Reduced phase reconstruction time to 5.5 ms, enabling 181 fps imaging.
- Achieved a mean phase reconstruction fidelity of up to 99.8%.
- Successfully demonstrated robust performance in experimental settings.
Conclusions:
- The learning-based MCF phase imaging method significantly accelerates QPI.
- This approach enables real-time, label-free endoscopic imaging in hard-to-reach areas.
- The open-source dataset facilitates further research and development in fiber-based QPI.
