Towards Autonomous Retinal Microsurgery Using RGB-D Images
Ji Woong Kim1, Shuwen Wei2, Peiyao Zhang1
1Mechanical Engineering Dept at the Johns Hopkins University, Baltimore, MD 21218 USA.
IEEE Robotics and Automation Letters
|September 23, 2024
Summary
This study presents an autonomous robotic system for subretinal injection (SI) in retinal surgery. The framework uses real-time OCT and microscope imaging with AI for precise, safe needle navigation, achieving 30 successful trials in pig eyes.
Area of Science:
- Ophthalmology
- Robotics
- Medical Imaging
Background:
- Retinal surgery demands extreme precision due to fragile tissue.
- Robotic assistance and advanced imaging like OCT enhance surgical capabilities.
- Subretinal injection (SI) requires accurate needle placement for drug delivery.
Purpose of the Study:
- To develop an autonomous framework for real-time needle navigation during subretinal injection (SI).
- To integrate optical coherence tomography (OCT) and microscope imaging with AI for automated surgical guidance.
- To enhance safety and performance in robotic-assisted retinal surgery.
Main Methods:
- Developed a system for autonomous needle navigation based on surgeon-defined waypoints.
- Integrated microscope and OCT imaging with Convolutional Neural Networks (CNNs) for segmentation.
- Employed model predictive control for trajectory generation respecting kinematic constraints.
Main Results:
- Successfully demonstrated 30 autonomous subretinal injection (SI) trials in pig eyes.
- The system achieved real-time navigation to the target subretinal space.
- Preliminary comparisons showed enhanced safety and performance over human-assisted robotic control.
Conclusions:
- The developed framework enables autonomous needle navigation for subretinal injection (SI).
- Integration of OCT, CNNs, and model predictive control offers a safe and effective approach.
- This technology has the potential to significantly improve outcomes in retinal surgery.
