股骨近端骨折发病率、致死率高，被称为“人生最后的骨折”。股骨近端骨折风险的准确预测具有显著的临床价值和社会意义。目前，临床上广泛使用骨密度（Bone Mineral Density, BMD）测定进行骨折风险预测，但预测结果并不理想。原因在于股骨近端骨折问题由松质骨质量、皮质骨质量、股骨颈干角、股骨颈长/径比、身高、体重等多种因素共同影响，并非单一的BMD所能决定。因此，发展更加科学、可靠、简便易用的股骨近端骨折风险综合预测方法迫在眉睫。要想解决这一问题，首先要面对松质骨非均质、非连续、各向异性的特点，其力学性能的反演是骨折风险准确评估的难点。

针对上述问题，本文以股骨近端骨折风险的定量预测为最终目标。首先，通过微结构分析和损伤参数反演，建立了基于临床CT的松质骨力学性能表征模型；之后以此为基础，建立了股骨近端骨折的力学模拟方法。进一步的，借助量纲分析和理论推导，确定了抗骨折强度与颈干角、股骨颈长/径比、股骨颈上后侧皮质厚度（S-P皮质厚度）以及Ward三角区CT值的定量关系。最后，通过抗骨折强度与个体预估侧落载荷的比较，建立了便于临床医生使用的股骨近端骨折风险综合性定量预测模型。主要研究内容及结论有：

（1）松质骨各向异性弹塑性参数表征。以人股骨头松质骨为研究对象，通过微结构的织构张量推导其弹性模量、剪切模量、屈服应力和剪切屈服应力等各向异性弹塑性力学参数，并通过最小二乘法拟合得到力学参数与CT值之间的经验公式。

（2）松质骨损伤参数反演。依据松质骨偏轴压缩试验规律，将松质骨材料本构简化为“三线性”模型，引入延性损伤模型描述其软化和破坏行为。建立偏轴压缩有限元模型，并反演极限强度、破坏起始应变与总失效位移等损伤参数。结果表明，松质骨具有较为稳定的破坏起始应变和总失效位移。屈服后强化模量约为弹性模量的5%，软化模量约为弹性模量的26%，极限应力约为屈服应力的102%，且比例系数的数值对CT值不敏感。这种不受CT值影响的稳定比例关系，可能是以应变表征的损伤参数较为稳定的内在原因。最后，选取部松质骨柱试样作为“验证集”，将经验公式计算所得的弹塑性参数以及反演所得损伤参数输入有限元模型，模拟松质柱单轴压缩，并与试验结果对比，验证了本文所提出松质骨力学性能表征模型的有效性。

（3）股骨近端骨折力学模拟方法建立。依据上述松质骨力学性能表征方法，针对最容易引发骨折的侧摔情况，建立了股骨近端骨折的有限元计算模型。通过不同CT值阈值区分Ward三角区和主要小梁束，按照其生理力线设置材料方向，解决了传统分区域赋值方法无法确定各向异性材料方向的问题。

（4）股骨近端骨折风险定量预测模型建立。将股骨近端简化为斜悬臂梁结构，借助量纲分析和理论推导，确定了抗骨折强度与CT图像上易获取的颈干角、股骨颈长细比、皮质厚度以及Ward三角CT值这几个参数的定量关系，通过抗骨折强度与个体预估侧落载荷的比较，建立了股骨近端骨折风险综合性定量预测公式。验证结果表明，在所涉及的7例样本范围内，本文所提出的预测公式能够正确筛选出7例骨折患者。

与基于骨密度测定的传统方法相比，本文所提出的骨折风险综合性定量评估公式能够依据CT图像中容易测量和获取的参数，帮助临床医生更加合理、精确、全面的评估患者股骨近端骨折风险，特别是在非骨质疏松但高骨折风险个体的识别和筛选方面具有显著优势。

Proximal femur fractures are associated with high morbidity and mortality and is known as the "last fracture of life". The accurate prediction of proximal femur fracture risk is of great clinical and social importance. Currently, Bone Mineral Density (BMD) is commonly used in clinical practice to predict fracture risk, but the results are not always satisfactory. The reason is that proximal femoral fracture risk is influenced by various factors such as cancellous bone quality, cortical bone quality, neck-shaft angle, neck slenderness ratio, height, weight, etc. Therefore, it is not all determined by single BMD parameter. To address this issue, the challenge caused by non-uniformity, discontinuity, and anisotropy of cancellous bone must be talked firstly, which means, accurately assessing fracture risk requires overcoming the difficulty of inverting its mechanical properties.

To address these problems, the aim of this study is the quantitative prediction of proximal femoral fracture risk. First, the model to characterize the mechanical properties of cancellous bone based on clinical CT was established by microstructural analysis and inversion of damage parameters. On this basis, the mechanical simulation method of proximal femur fracture was established. The quantitative relationship between fracture strength with neck-shaft angle, slenderness ratio, cortical thickness of the femoral neck and CT value in Ward's triangle was determined using dimensional analysis and theoretical derivation. Finally, a comprehensive quantitative prediction model for proximal femur fracture risk that is easy for clinicians to use was developed by comparing fracture strength with individual predicted side-fall load. The main research content and conclusions are:

(1) Characterization of anisotropic elastic-plastic parameters of cancellous bone. Focusing on human cancellous bone in the femoral head, the anisotropic elastic-plastic mechanical parameters were derived from the fabric tensor of the microstructure, including elastic modulus, shear elastic modulus, yield stress and shear yield stress. Then the empirical equations were established with respect to the CT values by least square fitting.

(2) Inversion of damage parameters of cancellous bone. Based on the law of uniaxial compression test of cancellous bone, the ontology of cancellous bone material was simplified into a 'tri-linear' model and a ductile damage model was introduced to describe its softening and destructive behavior. A finite element model for uniaxial compression was established. Damage parameters, including ultimate strength, fracture strain, and displacement at failure were inverted. The results indicated that cancellous bone exhibits a relatively stable fracture strain and failure displacement. Additionally, the post-yield strengthening modulus is about 5% of the elastic modulus, the softening modulus is about 26% of the elastic modulus, the ultimate stress is about 102% of the yield stress, and the scaling factors are insensitive to CT value. This stable proportionality may account for the greater stability of damage parameters characterized by strain.

(3) Establishment of mechanical simulation method for proximal femur fracture. With the foundation of the above cancellous bone mechanical parameter characterization method, a finite element model for proximal femur fracture for the case of sideways fall, which is the most likely cause of hip fracture, was established. Different CT value thresholds were used to distinguish the Ward’s triangle and main trabecular groups. Material direction was determined based on their natural orientation, resolving the issue of determining anisotropic material direction through traditional subregion assignment methods.

(4) The establishment of quantitative prediction model for proximal femur fracture risk. The proximal femur was simplified as an oblique cantilever beam structure, quantitative relationship between fracture load with neck-shaft angle, slenderness ratio, cortical thickness of the femoral neck and CT value in Ward's triangle was determined using dimensional analysis and theoretical derivation. Finally, a comprehensive quantitative prediction model for proximal femur fracture risk that is easy for clinicians to use was developed by comparing fracture strength with individual predicted side-fall load. The validation results indicated that using prediction formula proposed in this paper can correctly screen out 7 patients with fracture, within the range of samples included.

This paper proposed a comprehensive quantitative fracture risk assessment formula. Compared to the traditional method based on BMD, this formula allows a more rational, accurate and comprehensive assessment of proximal femoral fracture risk based on readily measurable parameters obtained from CT images. This is particularly useful in identifying and screening individuals without osteoporosis but at high risk of fracture, which is a significant advantage.