| 何子龙,廖昕,秦杰,曾辉,文婵娟,谢于雯,陈卫国.基于Logistic回归分析建立踝关节骨肿瘤良恶性鉴别简易评分模型及诊断效能[J].中国医学影像技术,2016,32(11):1735~1739 |
| 基于Logistic回归分析建立踝关节骨肿瘤良恶性鉴别简易评分模型及诊断效能 |
| Establishing simple scoring model based on Logistic regression analysis to identify benign and malignant bone tumor of ankle and its diagnosis effectiveness |
| 投稿时间：2016-05-09 修订日期：2016-08-03 |
| DOI：10.13929/j.1003-3289.2016.11.029 |
| 中文关键词: 踝关节 骨肿瘤 Logistic分析 评分模型 |
| 英文关键词:Ankle joint Bone neoplasms Logistic analysis Score model |
| 基金项目:广东省省级科技计划项目（2015B020233002、2015B020233008）、广东省科技计划项目（2016ZC0058）、广州市科技计划项目（201604020058）、南方医科大学大学生创新创业训练计划项目（201512121126）。 |
|
| 摘要点击次数: 1930 |
| 全文下载次数: 1020 |
| 中文摘要: |
| 目的 基于踝关节骨性结构（包括胫骨远端、腓骨远端及距骨）骨肿瘤的临床及X线特征建立评分模型，并评价其诊断效能。方法 回顾性分析接受X线检查并经病理证实的114例踝关节骨性结构骨肿瘤患者的临床与X线征象，纳入25个特征，临床9个 、X线16个 ，通过二分类Logistic回归分析，获取相关性高的征象，构建简易评分模型评价肿瘤良恶性，并利用ROC曲线分析该模型的诊断效能。结果 经二分类Logistic回归分析，年龄（OR=4.545）、病理性骨折（OR=2.567）、骨膜反应（OR=4.675）及软组织肿块（OR=8.148）为危险因素，长横径之比（OR=0.126）、硬化边（OR<0.001）、边缘（OR=0.074）及病灶主体（OR=0.070）为保护因素。建立简易评分模型：评分=（长横比+硬化边+边缘+病灶主体）-（年龄+病理骨折+骨膜反应+软组织肿块），以上各因素每存在一个累加1分，若无则为0分。此简易评分模型AUC为0.897（P<0.001），良恶性临界值为1.5分，敏感度为86.9%，特异度为74.7%。结论 踝关节简易评分系统有助于影像医师对踝关节骨性结构骨肿瘤定性诊断，可提高诊断效率并提供肿瘤良恶性鉴别的客观依据。 |
| 英文摘要: |
| Objective To establish simple scoring model to identify benign and malignant bone structure tumor of the ankle and diagnosis effectiveness based on the clinical and X-ray features. Methods The clinical feature and X-ray signs data of 114 cases of ankle bone structure tumor confirmed by pathology were retrospectively analyzed. Clinial feature included sex, age, accessible mass, tenderness, pain improves, get worse with exercise, increase skin temperature, night pain, movement disorder; X-ray included location (shins, fibulas and talus), quantity, long cross ratio (>1 cm or ≤1 cm), grown (expansive or infiltrative), calcifications, bony ridge, ground-glass opacity, sclerotic margins, edge, pathological fracture, articular surface, outside of bony cortex, the body of lesions, periosteal reaction, soft tissue mass, pseudarthrosis. High correlation signs were obtained by univariate and binary Logistic regression analysis, and a simple evaluation model was established, the diagnostic efficacy was evaluated with ROC curve. Results Through univariate and binary Logistic regression analysis, age (OR=4.545), pathological fracture (OR=2.567), periosteal reaction (OR=4.675) and soft tissue mass (OR=8.148) were risk factors; long cross ratio (OR=0.126), sclerotic margins (OR<0.001), the body of lesions (OR=0.070) and edge (OR=0.074) were protection factors. Score=(long cross ratio+sclerotic margins+close to the bone cortex+edge)-(age+pathological fracture+periosteal reaction+soft tissue mass), there was a cumulative one point when existed each of the above factors, whereas was 0 points when the absence. Taking 1.5 points as threshold, there was benign which was larger than 1.5 points, malignant which was less than 1.5 points, the sensitivity was 86.9%, specificity of 74.7%. Conclusion Ankle simple scoring model can help to qualitative ankle bone tumors, and improve the diagnostic efficiency and provide objective basis for benign and malignant identification. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
