股骨头头颈开窗治疗股骨头坏死预后的临床-影像组学分析  

Clinical-imaging analysis of the prognosis of femoral head and neck fenestration as the treatment for osteo‐necrosis of the femoral head

作　　者：席洪钟 陈浩 宋伟 刘成林 吴尧昆 付佳豪 孙光权 杜斌 刘锌 XI Hongzhong;CHEN Hao;SONG Wei;LIU Chenglin;WU Yaokun;FU Jiahao;SUN Guangquan;DU Bin;LIU Xin(Department of Orthopedics,Affiliated Hospital of Nanjing University of Chinese Medicine,Nanjing 210029,Jiangsu,China)

机构地区：[1]南京中医药大学附属医院骨伤科,南京210029

出　　处：《医学研究与战创伤救治》2025年第1期35-43,共9页Journal of Medical Research & Combat Trauma Care

基　　金：国家自然科学基金(82074471);江苏省研究生实践创新计划项目(SJCX23_0876)。

摘　　要：目的构建预测股骨头头颈开窗治疗股骨头坏死(ONFH)预后的临床-影像组学模型并探讨其预测价值。方法回顾性分析2010年6月至2021年6月南京中医药大学附属医院行股骨头头颈开窗手术的146例(169髋)ONFH患者的临床资料,根据纳入和排除标准最终共纳入106例患者(121髋)。按照随机抽样法(7∶3)将其随机分为训练集(84例)和验证集(37例)。比较保髋成功和失败患者的临床以及影像资料,构建临床、影像组学以及临床-影像组学模型。以ROC、决策曲线、校准曲线以及生存曲线评估模型的预测性能。结果年龄、JIC分型、术后免负重时间和术后持续接触危险因素是影响保髋结果的预测因素(P<0.05)。与头颈开窗打压植骨治疗ONFH预后显著相关的4个影像组学特征,包括原始形状伸长率、小波LLH一阶偏度、小波LHL glszm大面积低灰度加重以及小波HLL一阶中值。与影像组学模型和临床模型相比,训练集以及验证集中的临床-影像组学模型都表现出了更优越的预测性能。决策曲线显示,在高风险阈值的最大范围内,训练集以及验证集中使用临床-影像组学模型比临床模型以及影像组学模型方案获得了更多临床收益。在训练集中,JIC分期为C2型与最低的髋关节存活概率相关。此外,术后持续使用激素或酒精等风险因素也会显著降低髋关节存活概率。在训练集中,当Rad-score超过1.4089时,髋关节存活率明显降低。同样,对于年龄大于39岁、术后免负重时间少于5个月的患者,受累髋关节的存活率也相对较低。结论临床-影像组学模型在预测股骨头头颈开窗打压植骨治疗ONFH的预后方面表现良好,可辅助临床医师判断保髋患者的预后。Objective The study aims at developing a clinical-radiomic model for the prognosis of femoral head and neck fenestration as the treatment for osteonecrosis of the femoral head(ONFH),and exploring its predictive value.Methods The clinical data of 146 patients(169 hips)with ONFH who underwent femoral head and neck fenestration surgery in the Affiliated Hospital of Nanjing University of Chinese Medicine from June 2010 to June 2021 were retrospectively analyzed.According to the inclusion and exclusion criteria,a total of 106 patients(121 hips)were finally included.The patients were randomly divided into a training cohort(84 cases)and a validation cohort(37 cases)according to the random sampling method(7:3).Clinical and radiomic data of patients with successful and failed hip preservation were compared,and clinical,radiomic and clinical-radiomic models were established.The predictive performance of the models was assessed by ROC,decision curve,calibration curve,and survival curve.Results Age,JIC classification,postoperative non-weightbearing time,and continued postoperative exposure to risk factors were predictors of hip preservation outcomes(P<0.05).Four radiomic features significantly associated with the prognosis of ONFH treated with femoral head and neck fenestration,including original shape elongation,wavelet-LLH firstorder skewness,wavelet-LHL glszm large area low gray level emphasis,and wavelet-HLL firstorder median.Compared to the radiomic model and the clinical model in both the training cohort and the validation cohort,the clinical-radiomic model showed superior predictive performance.The decision curve showed that within the maximum range of the high-risk threshold,the use of clinical-radiomic model in the training cohort and in the validation cohort achieved more clinical benefits than the clinical model and the radiomic model.In the training cohort,JIC classification C2 was associated with the lowest probability of hip survival.In addition,risk factors such as continued use of glucocorticoids or alcohol a

关 键 词：股骨头坏死 预测模型 影像组学 股骨头头颈开窗 

分 类 号：R445[医药卫生—影像医学与核医学] R683[医药卫生—诊断学]