基于影像组学的老年髋部骨折风险评估

Hip fracture, with high disability rate and mortality, severely impaired the health status of the public. Previous studies were mostly about the bone mineral density (BMD) of femoral neck and total hip, which limited the ability of predicting hip fracture risks because only BMD of those regions couldn’t evaluate the whole bone status and density distribution. The present research is to perform computed tomography scanning of subjects with different genders and distinct age groups, and also hip fracture cases and controls. Radiomics is to be employed for achieving huge data amounts. The spatial distribution features of cancellous bone structure, the thickness of adjacent cortical bone, and the total volume of bone mass in proximal femur will be compared quantitatively to three-dimensionally visualize the variance between the subgroups by statistical parameter mapping. The correlation between quantitative indicators and gender, age, hip fractures should be comprehensively analyzed to explore the spatial distribution features of bone structure related to the osteoporotic hip fracture. The aim of our study is to find Radiomics features relatively sensitive and specific to the fractures and fracture high risk factors (gender, age) of the hip, build and validate hip fracture related Radiomics parameters and prediction models, and supply a new method for appropriately predicting and assessing the hip fracture risks in osteoporotic people.

老年髋部骨折的致残率及病死率极高，严重危害着人民的健康水平。而以往的研究多是针对股骨颈及全髋等几个区域骨密度进行研究，难以全面准确评估髋关节整体骨骼情况及密度分布，在预测髋部骨质疏松性骨折上存在较大局限性。本研究通过对不同性别、不同年龄段、髋部骨折与非骨折人群进行计算机断层扫描定量成像，利用基于影像组学（Radiomics）方法从影像中提取海量特征，高通量的获得骨松质与骨皮质的空间形态结果和矿盐含量的相关质量及空间数据信息，直观显示股骨近段各亚解剖区域骨质结构形态及骨量空间分布的变化规律，综合分析性别、年龄、骨折类型与各定量指标之间的相关性，旨在寻找和发现对髋部骨折及其高危因素（性别、年龄）相对敏感和特异的影像组学特征，构建并验证髋部脆性骨折相关的影像组学标签及预测模型，为在髋部骨折的高危人群中合理预测和评价骨折风险提供新的评估方法。

老年髋部骨折的致残率及病死率极高，严重危害着人民的健康水平。本研究对不同性别、不同年龄段、髋部骨折与非骨折人群进行计算机断层扫描定量成像，使用CT X-线吸收法（CT X-ray absorptiometry, CTXA）Hip软件分析系统、MIAF（Medical Image Analysis Framework）软件分析股骨近端各区域骨质情况，Osirix软件分析臀部、大腿中段和躯干肌肉情况，继而综合分析性别、年龄、骨折类型与各指标之间的相关性。对不同年龄段健康志愿者左髋部进行高分辨MRI扫描和有限元分析。本项目发现了髋关节股骨近段骨质结构形态空间分布特点随老龄化进展的演变规律；发现不同类型骨折患者在股骨近端不同区域的骨量分布差异；发现肌肉情况，特别是臀部肌肉密度是评估髋部脆性骨折的风险的一项重要指标，提出了包括骨密度和臀大肌、臀中小肌密度的骨折风险预测模型，具有重要意义。