Vision-based endoscope tracking for 3D ultrasound image-guided surgical navigation

Area of Science:

• Medical Imaging
• Robotics
• Surgical Navigation

Background:

• Endoscopic camera tracking is crucial for minimally invasive surgery.
• Conventional vision-based methods face challenges like initialization and scale ambiguity, especially in fetoscopic procedures.
• External tracking systems can be cumbersome and limit workflow integration.

Purpose of the Study:

• To develop a self-contained framework for endoscopic camera tracking.
• To integrate 3D ultrasonography with endoscopy for improved pose estimation.
• To overcome limitations of vision-based tracking in surgical applications.

Main Methods:

• A novel framework combining 3D ultrasonography and endoscopy was developed.
• Scene geometry from ultrasound was fused with endoscopic vision.
• The system was designed for seamless integration into surgical workflows without external trackers.

Main Results:

• The integrated approach successfully addressed initialization, scale ambiguity, and interest point inadequacy.
• Vision-based pose estimations were validated using phantom and ex vivo monkey placenta imaging.
• The framework demonstrated robust performance in complex scenarios.

Conclusions:

• The combined ultrasound-endoscopy approach offers a self-contained solution for accurate endoscopic camera tracking.
• This method enhances surgical navigation and has potential applications beyond fetoscopy, including general augmented reality in minimally invasive procedures.
• The system's ability to integrate without external devices simplifies surgical workflows.