Vision-based proximity detection in retinal surgery

Area of Science:

• Ophthalmology
• Medical Engineering
• Surgical Robotics

Background:

• Retinal surgery presents challenges like poor visualization, surgeon tremor, and lack of tactile feedback, increasing risks of iatrogenic retinal damage.
• Intraocular proximity sensing offers a promising solution to mitigate these limitations and improve surgical safety.

Purpose of the Study:

• To develop and evaluate a system for detecting unintentional collisions between surgical tools and the retina during ophthalmic surgery.
• To leverage visual feedback from stereo microscopes for enhanced intraocular proximity sensing.

Main Methods:

• A two-module system was developed: one for tracking surgical tool position in stereo images, and another for estimating retinal surface stereo disparity.
• The system was specifically designed to handle the difficult visualization conditions encountered in retinal surgery.
• Validation was performed using a silicon phantom eye and in vivo rabbit retinal data.

Main Results:

• The system successfully detected proximity between surgical tools and the retinal surface by analyzing stereo image disparity.
• The developed modules effectively operated under challenging visualization conditions typical of retinal procedures.
• Experimental testing demonstrated the system's potential clinical utility.

Conclusions:

• The proposed intraocular proximity sensing system can significantly enhance surgical safety in retinal procedures.
• Real-time detection of tool-retina proximity using stereo vision can prevent accidental surgical gestures and reduce retinal damage.
• This technology holds promise for improving outcomes in complex ophthalmic surgeries.