Toward Reliable AR-Guided Surgical Navigation: Interactive Deformation Modeling With Data-Driven Biomechanics and...

Area of Science:

• Medical Imaging
• Computer-Assisted Surgery
• Biomechanical Modeling

Background:

• Augmented reality (AR) surgical navigation relies on accurate preoperative organ models superimposed onto intraoperative anatomy.
• Maintaining alignment between preoperative models and dynamic patient anatomy is crucial for reliable AR guidance.
• Traditional finite element method (FEM) modeling is computationally expensive and struggles with significant anatomical changes during surgery.

Purpose of the Study:

• To develop a computationally efficient and accurate deformation modeling algorithm for AR-guided surgery.
• To integrate a human-in-the-loop mechanism for interactive surgeon correction of anatomical misalignments.
• To improve the reliability and safety of computer-assisted surgical navigation.

Main Methods:

• A novel data-driven biomechanics algorithm was developed to preserve FEM-level accuracy with improved computational efficiency.
• A human-in-the-loop mechanism was introduced, allowing surgeons to provide interactive prompts for real-time model correction.
• The algorithm was tested on a publicly available dataset for volumetric accuracy and registration error.

Main Results:

• The proposed algorithm achieved a mean target registration error of 3.42 mm.
• Incorporating surgeon prompts reduced the mean target registration error to 2.78 mm.
• The framework demonstrated superior volumetric accuracy compared to state-of-the-art methods.

Conclusions:

• The developed framework offers efficient and accurate deformation modeling for AR-guided surgery.
• Interactive surgeon collaboration enhances anatomical model adaptation and accuracy in complex surgical scenarios.
• This approach promises safer and more reliable computer-assisted surgical navigation.