Locally rigid, vessel-based registration for laparoscopic liver surgery

Area of Science:

• Medical Imaging
• Surgical Technology
• Computational Anatomy

Background:

• Laparoscopic liver resection offers benefits but faces challenges due to limited visualization and lack of tactile feedback.
• Image guidance is a potential solution, but organ deformation in soft tissues like the liver complicates registration.
• Accurate image guidance is crucial for overcoming the limitations of laparoscopic liver surgery.

Purpose of the Study:

• To propose and evaluate a laparoscopic ultrasound (LUS) image guidance system for liver surgery.
• To investigate the feasibility of locally rigid registration for aligning preoperative models with intraoperative ultrasound data.
• To assess the system's accuracy in a phantom and in vivo porcine models.

Main Methods:

• Developed a real-time segmentation method for extracting vessel center points from 2D LUS images.
• Utilized electromagnetically tracked LUS and preoperative CT-derived vessel centerline models.
• Employed landmark-based and iterative closest point (ICP) point-to-line registration techniques for aligning models.

Main Results:

• The locally rigid ICP method achieved an RMS residual error of 0.7 mm in phantom studies.
• Mean target registration error (TRE) in porcine studies was 3.58 mm and 2.99 mm.
• Landmark-based registration yielded mean TREs of 4.23 mm and 6.57 mm, depending on pneumoperitoneum conditions.

Conclusions:

• A practical image-guided surgery system using locally rigid registration of CT models to LUS-identified vasculature was developed.
• The system demonstrated accuracy suitable for surgical requirements in phantom and porcine liver resection models.
• Locally rigid registration is deemed sufficient for practical image guidance in laparoscopic liver surgery in the near future.