Brouwers Lars, Pull Ter Gunne Albert F, de Jongh Mariska A, Maal Thomas J J, Vreeken Rinaldo, van der Heijden Frank H W M, Leenen Luke P H, Spanjersberg Willem R, van Helden Sven H, Verbeek Diederik O, Bemelman Mike, Lansink Koen W W
Brabant Trauma Registry, Network Emergency Care Brabant, Elisabeth-Tweesteden Hospital, Tilburg, Noord-Brabant, The Netherlands.
Department of Surgery, RadboudUMC, Nijmegen, Gelderland, The Netherlands.
Eur J Orthop Surg Traumatol. 2020 Jan;30(1):109-116. doi: 10.1007/s00590-019-02537-w. Epub 2019 Sep 17.
Acetabular fractures are difficult to classify owing to the complex three-dimensional (3D) anatomy of the pelvis. 3D printing helps to understand and reliably classify acetabular fracture types. 3D-virtual reality (VR) may have comparable benefits. Our hypothesis is that 3D-VR is equivalent to 3D printing in understanding acetabular fracture patterns.
A total of 27 observers of various experience levels from several hospitals were requested to classify twenty 3D printed and VR models according to the Judet-Letournel classification. Additionally, surgeons were asked to state their preferred surgical approach and patient positioning. Time to classify each fracture type was recorded. The cases were randomized to rule out a learning curve. Inter-observer agreement was analyzed using Fleiss' kappa statistics (κ).
Inter-observer agreements varied by observer group and type of model used to classify the fracture: medical students: 3D print (κ = 0.61), VR (κ = 0.41); junior surgical residents: 3D print (0.51) VR (0.54); senior surgical residents: 3D print (0.66) VR (0.52); junior surgeons: 3D print (0.56), VR (0.43); senior surgeons: 3D print (κ = 0.59), VR (κ = 0.42). Using 3D printed models, there was more agreement on the surgical approach (junior surgeons κ = 0.23, senior surgeons κ = 0.31) when compared with VR (junior surgeons κ = 0.17, senior surgeons 0.25). No difference was found in time used to classify these fractures between 3D printing and VR for all groups (P = 1.000).
The Judet-Letournel acetabular classification stays difficult to interpret; only moderate kappa agreements were found. We found 3D-VR inferior to 3D printing in classifying acetabular fractures. Furthermore, the current 3D-VR technology is still not practical for intra-operative use.
由于骨盆复杂的三维（3D）解剖结构，髋臼骨折难以分类。3D打印有助于理解并可靠地分类髋臼骨折类型。3D虚拟现实（VR）可能有类似的益处。我们的假设是，在理解髋臼骨折模式方面，3D-VR等同于3D打印。
邀请来自几家医院的共27名不同经验水平的观察者，根据Judet-Letournel分类法对20个3D打印模型和VR模型进行分类。此外，要求外科医生说明他们偏好的手术入路和患者体位。记录对每种骨折类型进行分类的时间。病例随机分组以排除学习曲线的影响。使用Fleiss' kappa统计量（κ）分析观察者间的一致性。
观察者间的一致性因观察者组以及用于分类骨折的模型类型而异：医学生：3D打印（κ = 0.61），VR（κ = 0.41）；初级外科住院医师：3D打印（0.51），VR（0.54）；高级外科住院医师：3D打印（0.66），VR（0.52）；初级外科医生：3D打印（0.56），VR（0.43）；高级外科医生：3D打印（κ = 0.59），VR（κ = 0.42）。与VR（初级外科医生κ = 0.17，高级外科医生0.25）相比，使用3D打印模型时，在手术入路上的一致性更高（初级外科医生κ = 0.23，高级外科医生κ = 0.31）。对于所有组，3D打印和VR在对这些骨折进行分类时所用时间没有差异（P = 1.000）。
Judet-Letournel髋臼分类法仍然难以解读；仅发现中等程度的kappa一致性。我们发现3D-VR在髋臼骨折分类方面不如3D打印。此外，当前的3D-VR技术在术中使用仍不实用。
