Collaborative tracking for MRI-guided robotic intervention on the beating heart.

Area of Science:

• Medical Robotics and Interventional Imaging
• Cardiovascular Surgery and Intervention

Background:

• Current endoscopic procedures for aortic valve repair are time-consuming and costly.
• Real-time, robust tracking of anatomical landmarks is critical for robotic interventions in beating hearts.
• Maintaining precise control of interventional tools is essential to prevent damage during cardiac procedures.

Purpose of the Study:

• To develop and evaluate a real-time, robust tracking system for Magnetic Resonance Imaging (MRI)-guided robotic aortic valve repair.
• To enable precise adjustment of interventional tools using a closed feedback control loop.
• To assess the feasibility of the proposed tracking method in beating heart scenarios.

Main Methods:

• Implementation of a Bayesian network utilizing particle filter trackers for anatomical landmark point tracking.
• Development of a closed feedback control loop for continuous adjustment of the interventional tool.
• Testing the algorithm on 16 Magnetic Resonance Imaging (MRI) cine sequences.

Main Results:

• The Bayesian network particle filter tracker demonstrated real-time and robust performance.
• The algorithm proved flexible and general, adaptable to more complex tracking requirements.
• Experimental results confirmed the method's promise for MRI-guided robotic cardiac interventions.

Conclusions:

• The developed Bayesian network tracker effectively addresses the challenge of real-time landmark tracking in MRI-guided robotic aortic valve repair.
• This approach enhances the safety and precision of interventions on a beating heart.
• The method shows significant potential to reduce procedure time and cost compared to endoscopic guidance.