Assessing the precision of posttraumatic orbital reconstruction through “mirror” orbital superimposition : a novel approach for testing the anatomical accuracy

Orbital reconstruction in cases of trauma is usually performed using the unaffected side orbital volume as a reference, but this measurement does not fully consider the anatomical characteristics of orbital surfaces. We propose a novel procedure based on the registration of 3D orbital segmented surfaces. Reconstructed orbits from 20 patients and healthy orbits from 13 control subjects were segmented from the post-operative CT-scans. The 3D orbital model from the unaffected orbit was "mirrored" according to the sagittal plane and superimposed onto the reconstructed one, with calculation of volumes, asymmetry index and point-to-point RMS (root mean square) distances. Inter- and intra-observer errors were tested through Bland-Altman plot. Differences in volume, asymmetry index and RMS value between the control group and the treated patients were assessed through two-way ANOVA and Student's t-test (p < 0.05). According to Bland-Altman test, intra- and inter-operator repeatability was respectively 87% and 89%. No significant differences in volume or asymmetry index between the control group and the treated patients were observed (p > 0.05), but the RMS value was significantly larger in the latter ones (on average, 0.90 ± 0.26 mm vs. 0.67 ± 0.17 mm, p < 0.05). Results show that the reconstructed orbits present a morphologically different surface from the unaffected ones.