Related Experiment Video
Updated: Oct 25, 2025
05:12
Robotized Testing of Camera Positions to Determine Ideal Configuration for Stereo 3D Visualization of Open-Heart Surgery
Published on: August 12, 2021
2.2K
Robust Orthogonal-View 2-D/3-D Rigid Registration for Minimally Invasive Surgery
Zhou An1, Honghai Ma1, Lilu Liu2
1Department of Thoracic Surgery, The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou 310003, China.
Micromachines
|August 6, 2021
Summary
This study introduces a new deep learning framework for accurate 2-D/3-D rigid registration in minimally invasive surgery (MIS). The method enhances surgical robot guidance by improving intra-operative target pose estimation.
Area of Science:
- Medical Imaging
- Surgical Robotics
- Computer-Aided Surgery
Background:
- Accurate intra-operative target pose estimation is crucial for guiding surgical robots in minimally invasive surgery (MIS).
- Current 2-D/3-D rigid registration methods struggle to simultaneously achieve accuracy, robustness, and efficiency.
Purpose of the Study:
- To propose a novel orthogonal-view 2-D/3-D rigid registration framework for improved intra-operative target pose estimation.
- To enhance the accuracy and efficiency of aligning 2-D fluoroscopy with pre-operative 3-D CT data.
Main Methods:
- A deep learning-based dense reconstruction (X2CT-GAN) to generate a target CT from orthogonal fluoroscopy images.
- GPU-accelerated 3-D/3-D rigid registration for efficient alignment of reconstructed and pre-operative CT data.
Main Results:
- Achieved a mean target registration error (mTRE) of 1.65 ± 1.41 mm.
- Reported a gross failure rate (GFR) of 20% and a running time of 1.8 seconds.
- Outperformed state-of-the-art methods in most test cases.
Conclusions:
- The proposed framework offers a promising solution for accurate and efficient 2-D/3-D registration in MIS.
- This method has potential applications in localization and nano-manipulation for highly precise surgical interventions.
