Related Experiment Video
Updated: Sep 10, 2025
11:34
High-resolution, High-speed, Three-dimensional Video Imaging with Digital Fringe Projection Techniques
Published on: December 3, 2013
15.7K
Deep-learning-based endoscopic single-shot fringe projection profilometry
Ruizhi Zuo1, Shuwen Wei1, Yaning Wang1
1Johns Hopkins University, Department of Electrical and Computer Engineering, Baltimore, Maryland, United States.
Journal of Biomedical Optics
|August 21, 2025
Summary
A new single-shot fringe projection profilometry (FPP) system uses deep learning for real-time depth map generation. This endoscopic approach enhances surgical guidance accuracy and speed.
Area of Science:
- Medical Imaging
- Robotics
- Computer Vision
Background:
- Conventional fringe projection profilometry (FPP) is too slow for dynamic surgical measurements due to multiple image acquisitions.
- High-speed, accurate depth map generation is crucial for real-time robotic surgical guidance.
Purpose of the Study:
- To develop and demonstrate a deep-learning-based, single-shot FPP system for endoscopic surgical guidance.
- To achieve real-time, accurate depth map generation of target tissues.
Main Methods:
- An endoscopic single-shot FPP system using a dual-channel endoscope was designed.
- A deep learning network, combining MaskNet for segmentation and DepthNet for depth prediction, was developed.
- A data synthesis method was employed to create diverse training datasets.
Main Results:
- The system achieved a maximum depth prediction error of approximately 2 mm.
- Processing time per frame was around 12.75 ms.
- An optimal fringe pattern frequency of 20 Hz was identified for the single-shot FPP setup.
Conclusions:
- The deep-learning-based single-shot FPP endoscopic system effectively generates real-time depth maps with millimeter-scale accuracy.
- This technology has the potential to significantly improve the reliability of image-guided robotic surgery.
