Augmented Reality-Based Femur Registration With Head-Mounted Display and Infrared Tracking Device as Stand-Alone Navigation Tool: A Preclinical Validation Study

Abstract

Background and Aims This study evaluated the accuracy of a standalone augmented reality (AR) navigation system using the Microsoft HoloLens 2 for total hip arthroplasty (THA). The system integrates infrared (IR) tracking and a preoperative 3D CT model for intraoperative visualization and guidance. The aim was to assess whether this AR-based solution could achieve clinically acceptable accuracy for use as a surgical navigation tool.

Methods A dedicated AR application was developed to perform six-degrees-of-freedom pose estimation using inside-out tracking. Registration accuracy was assessed using six femur replicas with predefined target points. The registration pipeline combined landmark-based initialization with iterative surface refinement. Target registration error (TRE) was calculated for each point, and the influence of anatomical region, axis, surgical approach (anterior vs. posterior), and user experience level was analyzed.

Results The system achieved a mean TRE of 3.61 ± 2.18 mm. Significant variations in accuracy were observed between anatomical regions (p = 0.019) and along different axes (p < 0.001), with the highest errors noted along the anteroposterior axis and in distal femoral regions. No significant differences were found between anterior and posterior approaches or among users with varying levels of experience, indicating operator-independent performance.

Conclusion The AR navigation system demonstrated consistent and accurate registration performance across users and approaches. With a mean TRE of ~3 mm, the system meets accuracy requirements for potential clinical application in THA. Further cadaveric validation is recommended to confirm surgical feasibility and applicability.