How Accurate Is the Treatment of Midfacial Fractures by a Specific Navigation System Integrating “Mirroring” Computational Planning? Beyond Mere Average Difference Analysis

Abstract

To evaluate the accuracy of a specific navigation system integrating "mirroring" computational planning in the treatment of midfacial fractures by comparing planned with actual postoperative 3-dimensional (3D) images. Preoperative planned and postoperative 3D computed tomographic (CT) and cone-beam CT (CBCT) images of 20 patients with midfacial fractures were analyzed. Images were fused using dedicated software (iPlan Cranial 2.6). They were imported in Standard Tessellation Language (STL) format to open-source software (Meshlab) and resized to delineate the surgically repositioned portion of bone. The images were imported in STL format to ad hoc software for calculating the surface differences between the 2 3D images. The distribution of the differences was assessed using boxplots for each patient, and the proportion of the actual image close to the planned image was estimated by the proportion of values within an accuracy cutoff set at ±2 mm. The mean difference between the 2 3D surfaces was 0.12 mm. The proportion of values between the 2 surfaces and included within the interval of accuracy was greater than 90% in 6 patients (30%), 80 to 90% in 6 patients (30%), 50 to 80% in 7 patients (35%), and less than 50% in 1 patient (5%). This study showed that post-traumatic midfacial reconstruction can be accurately approximated and thus predicted by a specific navigation system integrating "mirroring" computational planning for most patients. Further study should examine risk factors for inaccurate prediction.