Wu Shenghui, Wang Wei, Li Ruiyang, Guo Jingyi, Miao Yu, Li Guangyi, Mei Jiong
Department of Orthopedic Surgery, Shanghai Sixth People's Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China.
Department of Biomedical Engineering, The Hong Kong Polytechnic University, Hong Kong, China.
Front Bioeng Biotechnol. 2023 Jan 17;11:1115639. doi: 10.3389/fbioe.2023.1115639. eCollection 2023.
The injury of femoral head varies among femoral head fractures (FHFs). In addition, the injury degree of the femoral head is a significant predictor of femoral neck fracture (FNF) incidence in patients with FHFs. However, the exact measurement methods have yet been clearly defined based on injury models of FHFs. This study aimed to design a new measurement for the injury degree of the femoral head on 2D and 3D models with computed tomography (CT) images and investigate its association with FHFs with FNF. A consecutive series of 209 patients with FHFs was assessed regarding patient characteristics, CT images, and rate of FNF. New parameters for injury degree of femoral head, including percentage of maximum defect length (PMDL) in the 2D CT model and percentage of fracture area (PFA) in the 3D CT-reconstruction model, were respectively measured. Four 2D parameters included PMDLs in the coronal, cross-sectional and sagittal plane and average PMDL across all three planes. Reliability tests for all parameters were evaluated in 100 randomly selected patients. The PMDL with better reliability and areas under curves (AUCs) was finally defined as the 2D parameter. Factors associated with FNF were determined by binary logistic regression analysis. The sensitivity, specificity, likelihood ratios, and positive and negative predictive values for different cut-off values of the 2D and 3D parameters were employed to test the diagnostic accuracy for FNF prediction. Intra- and inter-class coefficients for all parameters were ≥0.887. AUCs of all parameters ranged from 0.719 to 0.929 ( < 0.05). The average PMDL across all three planes was defined as the 2D parameter. The results of logistic regression analysis showed that average PMDL across all three planes and PFA were the significant predictors of FNF ( < 0.05). The cutoff values of the average PMDL across all three planes and PFA were 91.65% and 29.68%. The sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, predictive positive value and negative predictive value of 2D (3D) parameters were 91.7% (83.3%), 93.4% (58.4%), 13.8 (2.0), 0.09 (0.29), 45.83% (10.87%), and 99.46% (98.29%). The new measurement on 2D and 3D injury models with CT has been established to assess the fracture risk of femoral neck in patients with FHFs in the clinic practice. 2D and 3D parameters in FHFs were a feasible adjunctive diagnostic tool in identifying FNFs. In addition, this finding might also provide a theoretic basis for the investigation of the convenient digital-model in complex injury analysis.
股骨头损伤在股骨头骨折（FHFs）中各不相同。此外，股骨头的损伤程度是FHFs患者发生股骨颈骨折（FNF）的重要预测指标。然而，基于FHFs损伤模型的确切测量方法尚未明确界定。本研究旨在利用计算机断层扫描（CT）图像在二维和三维模型上设计一种新的股骨头损伤程度测量方法，并研究其与伴有FNF的FHFs之间的关联。对连续的209例FHFs患者进行了患者特征、CT图像和FNF发生率的评估。分别测量了股骨头损伤程度的新参数，包括二维CT模型中的最大缺损长度百分比（PMDL）和三维CT重建模型中的骨折面积百分比（PFA）。四个二维参数包括冠状面、横断面和矢状面的PMDL以及所有三个平面的平均PMDL。在100例随机选择的患者中对所有参数进行了可靠性测试。最终将可靠性更好且曲线下面积（AUCs）的PMDL定义为二维参数。通过二元逻辑回归分析确定与FNF相关的因素。采用二维和三维参数不同截断值的敏感性、特异性、似然比以及阳性和阴性预测值来测试FNF预测的诊断准确性。所有参数的组内和组间系数均≥0.887。所有参数的AUCs范围为0.719至0.929（<0.05）。所有三个平面的平均PMDL被定义为二维参数。逻辑回归分析结果表明，所有三个平面的平均PMDL和PFA是FNF的重要预测指标（<0.05）。所有三个平面的平均PMDL和PFA的截断值分别为91.65%和29.68%。二维（三维）参数的敏感性、特异性、阳性似然比、阴性似然比、预测阳性值和阴性预测值分别为91.7%（83.3%）、93.4%（58.4%）、13.8（2.0）、0.09（0.29）、45.83%（10.87%）和99.46%（98.29%）。已建立基于CT的二维和三维损伤模型的新测量方法，以评估临床实践中FHFs患者股骨颈的骨折风险。FHFs中的二维和三维参数是识别FNFs的一种可行的辅助诊断工具。此外，这一发现也可能为复杂损伤分析中便捷数字模型的研究提供理论基础。
