Computer-Aided Breast Surgery Framework Using a Markerless Augmented Reality Method.

Area of Science:

• Medical Imaging and Computer-Assisted Surgery
• Augmented Reality in Healthcare
• Surgical Navigation Systems

Background:

• Accurate pre-operative imaging is crucial for effective breast lesion removal.
• Marker-based registration methods in AR surgery can be cumbersome and introduce artifacts.
• Integrating 3D patient-specific data into real-time AR visualization remains a challenge.

Purpose of the Study:

• To develop and validate a markerless Augmented Reality (AR) surgical framework for breast lesion removal.
• To enable precise registration of 3D breast Computed Tomography (CT) data onto the patient's anatomy using a depth sensor.
• To assess the accuracy of AR visualization for surgical guidance without external markers.

Main Methods:

• Acquisition of a patient mesh in the real coordinate system using a depth sensor.
• Generation of a virtual coordinate system patient mesh via contrast-based skin segmentation from 3D CT data.
• Nipple detection for Region of Interest (ROI) definition, followed by center of mass alignment and Iterative Closest Point (ICP) registration.

Main Results:

• Achieved 98.35 ± 0.71% skin segmentation accuracy (Dice Similarity Coefficient).
• Demonstrated a nipple detection error of 2.79 ± 1.54 mm.
• Reported a registration error of 4.69 ± 1.95 mm, confirming high accuracy in AR visualization.

Conclusions:

• The proposed markerless AR framework enables accurate 3D visualization of medical data on the patient's body.
• A single depth sensor is sufficient for registration, eliminating the need for markers.
• This approach holds significant potential for improving precision in breast lesion removal surgery.