Physics-based shape matching for intraoperative image guidance.

Area of Science:

• Medical physics
• Computer-assisted surgery
• Image-guided interventions

Background:

• Soft-tissue deformations challenge preoperative planning in computer-assisted interventions.
• Intraoperative imaging is crucial for compensating tissue movement.
• Accurate surface matching of intraoperative data to preoperative models is essential but difficult due to sensor noise and partial visibility.

Purpose of the Study:

• To propose a novel physics-based shape matching (PBSM) approach for registering intraoperative surface meshes to preoperative data.
• To address the challenges of nonrigid registration in real-time surgical settings.
• To enable accurate compensation for soft-tissue deformations during interventions.

Main Methods:

• Developed a physics-based shape matching (PBSM) method modeling registration as an electrostatic-elastic problem.
• Utilized the finite element (FE) method for efficient energy functional solving.
• Integrated PBSM with rigid registration and landmark-based methods for robust performance.
• Applied the approach to image guidance in laparoscopic interventions.

Main Results:

• PBSM demonstrated robustness to initial registration and parameter variations in simulations.
• Achieved submillimeter registration accuracy (0.32–0.46 mm mean error).
• An unoptimized implementation showed near real-time performance (7–19 s).
• Successfully matched partial surfaces and demonstrated application in laparoscopic phantom experiments.

Conclusions:

• The PBSM approach is fast, robust, and accurate for soft-tissue surface matching.
• Leverages preoperative FE models to recover volumetric structures like tumors and vessels.
• Applicable to laparoscopic surgery and potentially other areas requiring sparse sensor data registration, such as radiation therapy.