Comparative analysis between optical see-through and video see-through technologies: usability and accuracy assessment of two head-mounted displays for cranio-maxillofacial surgery

In recent years, Mixed Reality (MR) technologies in surgery have attracted growing interest. To assess the usability and accuracy of MR in surgery, it is essential to compare the devices available on the market. However, in the literature, no studies compare optical see-through (OST) and video see-through (VST) head-mounted displays for surgical applications, considering both accuracy and usability. The paper aims to compare Microsoft HoloLens 2 (OST) and Varjo XR-3 (VST) in the context of Cranio-Maxillofacial surgery. The analysis is based on an MR application designed to assist surgeons during two critical phases of cranio-maxillofacial procedures: osteotomy and repositioning. The application renders cutting trajectories and drilling points as emissive-shaded pipe geometries. It provides real-time positional feedback for osseous fragments, using cubic fiducial markers to ensure robust spatial registration.Quantitative accuracy was assessed by measuring deviations between nominal and user-drawn osteotomy geometries on 3D-printed phantoms, and by evaluating translational and angular errors during maxillary fragment repositioning. Qualitative evaluation employed the User Experience Questionnaire (UEQ), a structured UX survey, and a post-test interview, administered to 9 participants (3 clinicians and 6 engineers).Results indicate that the HoloLens 2 achieves sub-millimetric mean errors (< 1 mm) during osteotomy guidance, lower than those observed with the Varjo XR-3 (mean 1.08 mm). In contrast, the Varjo XR-3 outperforms in fragment repositioning (0.84 mm vs. 1.41 mm). Qualitatively, users favored the Varjo system for hologram clarity and menu interaction, while the HoloLens 2 was preferred for its lighter weight, untethered operation, and superior spatial awareness in prolonged use. These findings suggest that OST devices may be optimal for precise osteotomy guidance, whereas VST systems excel in tasks requiring high-fidelity fragment alignment.