Abstract
PURPOSE: Consumer-grade virtual reality (VR) has recently enabled various medical applications, but more evidence supporting their validity is needed. We investigated the accuracy of simulated surgical planning in a VR environment (VR) with temporal bones and compared it to conventional cross-sectional image viewing in picture archiving and communication system (PACS) interface. METHODS: Five experienced otologic surgeons measured significant anatomic structures and fiducials on five fresh-frozen cadaveric temporal bones in VR and cross-sectional viewing. Primary image data were acquired by computed tomography. In total, 275 anatomical landmark measurements and 250 measurements of the distance between fiducials were obtained with both methods. Distance measurements between the fiducials were confirmed by physical measurement obtained by Vernier caliper. The experts evaluated the subjective validity of both methods on a 5-point Likert scale qualitative survey. RESULTS: A strong correlation based on intraclass coefficient was found between the methods on both the anatomical (r > 0.900) and fiducial measurements (r > 0.916). Two-tailed paired t-test and Bland-Altman plots demonstrated high equivalences between the VR and cross-sectional viewing with mean differences of 1.9% (p = 0.396) and 0.472 mm (p = 0.065) for anatomical and fiducial measurements, respectively. Gross measurement errors due to the misidentification of fiducials occurred more frequently in the cross-sectional viewing. The mean face and content validity rating for VR were significantly better compared to cross-sectional viewing (total mean score 4.11 vs 3.39, p < 0.001). CONCLUSION: Our study supports good accuracy and reliability of VR environment for simulated surgical planning in temporal bones compared to conventional cross-sectional visualization.