Page 60 - Advanced concepts in orbital wall fractures

58
Chapter 3
There was no relevant difference in symmetry between men and women. The absolute volumetric difference and standard deviation were comparable (p=0.59) in men and women. The only notable and expected difference was that, on average, men had markedly larger bony orbits than women.
In addition to asymmetry in orbital volume, the orbital contour was studied. To achieve a pretraumatised anatomy after orbital reconstruction, it is important to restore the contour for correct support of the soft tissue. The thin shell structure of the orbital floor and medial wall consists of many curves and angles, where left-versus-right differences might exist29. The absolute mean difference in the study population was 0.83 mm (95th percentile, 1.91 mm). This proves that differences do occur in the anatomy of the orbital walls. Inspection of the distance maps showed that variations in symmetry are not centred on a specific part of the orbital wall, but rather are spread out across the entire orbital cavity. However, when analysing the individual distance maps, the apex usually showed differences.
Previous studies have presented left-versus-right differences of up to 8 % in volume (average up to 3.5 %; Table 3)30-32. In the authors’ opinion, this might not be representative for a normal human population. The results could be caused by inaccuracy of the measuring technique or the small study. Kamer et al. stated that there is a major interindividual shape and size variability in the orbital floor33. In another study on interindividual morphologic differences, Kamer et al. stated that there was no relevant volume laterality between left and right orbits, which is comparable to the present results34. In 14 % of patients, left and right differences were larger than 1 mL. This is greater than the 4 % found in the present study. In general, the results of this study seem to agree with the literature for contour and volume.
In this study, no manual segmentation of the orbital volume was performed, because this is labour intensive and time-consuming. Instead, a semi-automatic software segmentation method was used. This method was previously validated and was considered accurate, reproducible, and efficient20.