Robust feature tracking on the beating heart for a robotic-guided endoscope.

Area of Science:

• Medical Robotics
• Surgical Visualization
• Computer Vision

Background:

• Minimally invasive bypass surgery requires enhanced visualization of the beating heart.
• Robotic-guided endoscopes and visual servoing can improve surgical views.
• Accurate, real-time feature tracking on the heart is crucial for these technologies.

Purpose of the Study:

• To develop and test marker-less algorithms for accurate, robust, and real-time motion tracking of features on a beating heart using an uncalibrated endoscope.
• To evaluate the performance of different tracking algorithms under various conditions.

Main Methods:

• Comparison of Lucas-Kanade pyramidal optical flow and Speeded-Up Robust Features (SURF)-based methods.
• Evaluation using metrics for accuracy, robustness, and drift.
• Experiments included static and moving endoscopes, and a visual servoing application.

Main Results:

• The combination of Lucas-Kanade tracking and SURF feature detection demonstrated superior accuracy and robustness.
• Optimal parameters were identified: 51x51 pixel window size and 20-pixel interframe motion threshold.
• Successful integration of feature tracking into uncalibrated visual servoing with a robotic endoscope.

Conclusions:

• Real-time, robust feature tracking on beating hearts from endoscopic video is achievable.
• This technology can significantly aid robotically-assisted minimally invasive bypass surgery and conventional laparoscopic procedures.