EasyREG: Easy Depth-Based Markerless Registration and Tracking using Augmented Reality Device for Surgical Guidance.

Area of Science:

• Medical Technology
• Computer-Aided Surgery
• Augmented Reality

Background:

• Traditional surgical guidance relies on fiducial markers, posing clinical challenges.
• Existing markerless AR solutions struggle with accuracy due to occlusions and sensor data outliers.

Purpose of the Study:

• To develop a robust, markerless framework for precise surgical guidance using AR devices.
• To overcome limitations of current marker-based and markerless tracking systems in medical applications.

Main Methods:

• A two-module system: registration for accurate anatomy localization and tracking for real-time pose estimation.
• Utilizes AR depth sensors, depth error correction, human-in-the-loop filtering, and curvature-aware feature sampling.
• Incorporates Iterative Closest Point (ICP) refinement for precise alignment.

Main Results:

• The markerless system demonstrates superior registration accuracy compared to industrial solutions.
• Tracking performance is comparable to existing commercial systems.
• The framework effectively handles both static and moving target anatomy.

Conclusions:

• The proposed markerless framework offers a high-precision, outlier-robust solution for surgical guidance.
• It provides a comprehensive, one-stop system for diverse surgical scenarios.
• This technology has the potential to improve the safety and efficiency of AR-assisted surgery.