Intelligent viewpoint selection for efficient CT to video registration in laparoscopic liver surgery.

Area of Science:

• Medical Imaging
• Surgical Technology
• Computer-Assisted Surgery

Background:

• Minimally invasive surgery (MIS) offers patient benefits but faces challenges like limited tactile feedback and bleeding control.
• Augmented reality (AR) can improve MIS by fusing pre-operative CT data with real-time laparoscopic video for better visualization.
• Accurate and fast video-to-CT registration is crucial for effective AR visualization during surgery.

Purpose of the Study:

• To develop and validate a view planning method for efficient rigid registration in laparoscopic surgery.
• To improve the accuracy and speed of video-to-CT registration for augmented reality applications.
• To reduce interruptions during surgical procedures by optimizing data acquisition for AR visualization.

Main Methods:

• Proposed a view planning approach for efficient rigid registration based on trocar position.
• Sampled and scored camera positions using liver surface properties.
• Implemented a simulation framework with a segmented CT model and validated on clinical data from liver resection.

Main Results:

• Demonstrated a viewpoint scoring strategy and identified reliable liver regions for accurate registration.
• Showcased wider convergence basins for higher-scoring viewpoints.
• Achieved comparable registration performance using fewer, high-scoring views versus multiple low-scoring views, suggesting a shift towards distinctive patch acquisition for surface reconstruction.

Conclusions:

• The simulation framework shows promising results for view planning in laparoscopic liver surgery.
• Further research into comprehensive view planning methods is motivated for efficient registration.
• The proposed method offers a more explicit protocol for surface reconstruction in AR-assisted surgery.