Augmented reality as an aid in maxillofacial surgery: Validation of a wearable system allowing maxillary repositioning

Aim We present a newly designed, localiser-free, head-mounted system featuring augmented reality as an aid to maxillofacial bone surgery, and assess the potential utility of the device by conducting a feasibility study and validation. Methods Our head-mounted wearable system facilitating augmented surgery was developed as a stand-alone, video-based, see-through device in which the visual features were adapted to facilitate maxillofacial bone surgery. We implement a strategy designed to present augmented reality information to the operating surgeon. LeFort1 osteotomy was chosen as the test procedure. The system is designed to exhibit virtual planning overlaying the details of a real patient. We implemented a method allowing performance of waferless, augmented-reality assisted bone repositioning. In vitro testing was conducted on a physical replica of a human skull, and the augmented reality system was used to perform LeFort1 maxillary repositioning. Surgical accuracy was measured with the aid of an optical navigation system that recorded the coordinates of three reference points (located in anterior, posterior right, and posterior left positions) on the repositioned maxilla. The outcomes were compared with those expected to be achievable in a three-dimensional environment. Data were derived using three levels of surgical planning, of increasing complexity, and for nine different operators with varying levels of surgical skill. Conclusion We used a new localiser-free, head-mounted, wearable, stereoscopic, video see-through display to develop a useful strategy affording surgeons access to augmented reality information. Our device appears to be accurate when used to assist in waferless maxillary repositioning. Our results suggest that the method can potentially be extended for use with many surgical procedures on the facial skeleton. Further, our positive results suggest that it would be appropriate to proceed to in vivo testing to assess surgical accuracy under real clinical conditions. Results The mean error was 1.70 ± 0.51 mm. The axial errors were 0.89 ± 0.54 mm on the sagittal axis, 0.60 ± 0.20 mm on the frontal axis, and 1.06 ± 0.40 mm on the craniocaudal axis. The simplest plan was associated with a slightly lower mean error (1.58 ± 0.37 mm) compared with the more complex plans (medium: 1.82 ± 0.71 mm; difficult: 1.70 ± 0.45 mm). The mean error for the anterior reference point was lower (1.33 ± 0.58 mm) than those for both the posterior right (1.72 ± 0.24 mm) and posterior left points (2.05 ± 0.47 mm). No significant difference in terms of error was noticed among operators, despite variations in surgical experience. Feedback from surgeons was acceptable; all tests were completed within 15 min and the tool was considered to be both comfortable and usable in practice