Soft tissue motion tracking with application to tablet-based incision planning in laser surgery

Area of Science:

• Computer-assisted surgery
• Medical image analysis
• Surgical robotics

Background:

• Laser surgery incision planning using stylus and tablet interfaces shows promise, outperforming micromanipulator control.
• Adapting to dynamic surgical scenes using vision-based methods remains a challenge.
• Real-time motion compensation is crucial for enhancing precision in image-guided interventions.

Purpose of the Study:

• To develop and evaluate a vision-based scene motion compensation method for tablet-based laser surgery planning.
• To address the challenge of dynamic tissue deformation tracking in real-time.
• To improve the accuracy and usability of surgical planning systems through enhanced image stabilization.

Main Methods:

• A stereo-based motion tracking system utilizing piecewise affine deformation modeling.
• Incorporation of epipolar constraints for left-right consistency in energy minimization.
• Implementation of illumination-invariant tracking and appearance-based occlusion detection.
• Validation on laparoscopic and laryngeal in vivo data, including simulated laser cuttings and occlusions.

Main Results:

• Achieved root-mean-square error of 2.45 mm (laparoscopic) and 0.41 mm (laryngeal).
• Demonstrated robust stereoscopic tracking under varying illumination, translation, rotation, and scale.
• Occlusion detection significantly improved robustness against tracking failures.
• User study confirmed increased path tracing accuracy and usability with the integrated tracking for image stabilization.

Conclusions:

• The developed algorithm successfully extends piecewise affine deformation tracking to stereo vision, incorporating epipolar constraints.
• The method enhances surgical performance in laser incision planning, demonstrating its potential for broader computer-assisted surgery applications.
• Real-time tissue deformation tracking and image stabilization are key to improving surgical precision and user experience.