Mixed reality for spine surgery: a step into the future with a human cadaveric accuracy study.

Abstract

Current application of mixed reality as a navigation aid in the field of spinal navigation points to the potential of this technology in spine surgery. Crucial factors for acceptance include intuitive workflow, system stability, reliability, and accuracy of the method. The authors therefore aimed to investigate the accuracy of the system in visualization of anatomical structures using mixed reality in the example of pedicles of the thoracic spine in a human cadaveric study. Potential difficulties and limitations are discussed. CT scans of a human cadaveric spinal column specimen were performed. After segmentation and import into the advanced HoloLens 2 software, the vertebrae were exposed. The vertebral arches were preserved on one side for a landmark-based surface registration, whereas pedicles were exposed on the other side in order to measure and evaluate deviation of the overlay holographs with regard to the exact anatomical structure. Accuracy was measured and statistically evaluated. In this work it was demonstrated that the overlay of the virtual 3D model pedicles with the real anatomical structures with anatomical landmark registration was within an acceptable surgical accuracy with the mean value of 2.1 mm (maximum 3.8 mm, minimum 1.2 mm). The highest accuracy was registered at the medial and lateral pedicle wall, and the measurement results were best in the region of the middle thoracic spine. The accuracy analysis for mixed reality (i.e., between the virtual and real anatomical situation of the thoracic spine) showed a very good agreement when focus was on the pedicles. This work is thus a rare proof of the precision of segmentation to the potential surgical area. The results encourage researchers to open up mixed reality technology in its development and application for spinal navigation.