Augmented reality calibration using feature triangulation iteration-based registration for surgical navigation.

Area of Science:

• Medical Imaging
• Computer-Assisted Surgery
• Augmented Reality

Background:

• Marker-based augmented reality (AR) calibration for surgical navigation often requires additional CT scans, increasing costs and limiting clinical use.
• Existing methods suffer from high manufacturing costs and insufficient accuracy for widespread adoption.

Purpose of the Study:

• To introduce a novel AR calibration framework for surgical navigation that enhances accuracy and reduces reliance on multiple scans.
• To develop an efficient and precise method for calibrating the spatial relationship between virtual and real objects in surgical navigation.

Main Methods:

• A framework combining a Microsoft HoloLens with a single camera registration module was developed.
• Multi-view patient images were captured for reconstruction, with shape feature matching for model size adjustment.
• 3D point cloud segmentation and 3D line segment detection were used to extract image marker corner points for registration.
• A feature triangulation iteration method enabled accurate pose calibration between the image marker and the patient.

Main Results:

• The proposed AR calibration method achieved average fusion errors of 0.70 ± 0.16 mm (phantoms) and 0.91 ± 0.13 mm (volunteers).
• The method demonstrated superior fusion accuracy compared to six advanced AR calibration techniques.
• Clinical feasibility was confirmed through a simulated liver puncture experiment on a volunteer.

Conclusions:

• The developed AR calibration method is effective and shows significant potential for enhancing surgical navigation and performance.
• This approach offers a more accurate and potentially cost-effective solution for AR-assisted surgical procedures.