基于CT影像学特征构建原发性小肠淋巴瘤与间质瘤的鉴别诊断预测模型  

Construct predictive model in the differential diagnosis of small intestinal lymphoma and stromal tumor

作　　者：陈春茂 孙西河 CHEN Chunmao;SUN Xihe(Medical Imaging College of Shandong Second Medical University,Weifang 261053,China;Imaging Center of Shouguang People's Hospital;Imaging Center,the Affiliated Hospital of Shandong Second Medical University)

机构地区：[1]山东第二医科大学医学影像学院,山东潍坊261053 [2]寿光市人民医院影像中心 [3]山东第二医科大学附属医院影像中心

出　　处：《潍坊医学院学报》2024年第3期161-164,共4页Acta Academiae Medicinae Weifang

基　　金：山东省自然科学基金面上项目(项目编号:ZR2017MH110)。

摘　　要：目的 探讨基于CT影像学特征构建的预测模型对原发性小肠淋巴瘤与间质瘤的鉴别诊断价值。方法 选取经病理证实的原发性小肠淋巴瘤、间质瘤患者各50例,比较两组患者临床资料、CT影像学特征,采用二元Logistic回归分析等筛选出有鉴别意义的独立影响因素并构建鉴别诊断预测模型,绘制受试者工作特征(ROC)曲线,评价该预测模型的鉴别诊断价值。结果 两组患者动脉期净强化CT值、病灶周围肿大淋巴结、血管漂浮征、病理血管比较,差异均有统计学意义(P<0.05),均是具有鉴别意义的独立影响因素(P<0.05),基于这4个CT影像学特征构建鉴别诊断预测模型,ROC曲线显示其AUC为0.971(95%CI:0.941-1.000),敏感度为0.92,特异度为0.96.结论 基于CT影像学特征构建的预测模型对小肠淋巴瘤与间质瘤的鉴别诊断具有较高应用价值。Objective To explore the value of a prediction model based on CT imaging characteristics in differential diagnosis of primary small intestinal lymphoma and stromal tumor.Methods Fifty cases of small intestinal lymphoma and 50 cases of stromal tumor confirmed by pathology were collected respectively,and the clinical data and CT imaging characteristics of the two groups were compared.The independent influencing factors with differential signifi-cance for the two groups were screened by binary Logistic regression analysis,and a differential diagnosis prediction mod-el was constructed,and the working characteristics(ROC)curve of the subjects was drawn to evaluate the differential di-agnosis value of the prediction model.Results The difference between the two groups was statistically significant(P<0.05),including the value of net enhanced CT in arterial phase,swollen lymph nodes around the lesion,vascular floating sign and pathological blood vessels,which were all independent influencing factors with differential significance(P<0.05).Based on these four CT imaging features,a differential diagnosis prediction model was constructed,and the ROC curve showed that its AUC was 0.971(95%CI:0.941-1.000).Conclusion The prediction model based on CT imaging features has high application value in the differential diagnosis of small intestinal lymphoma and stromal tumor.

关 键 词：原发性小肠淋巴瘤 胃肠道间质瘤 鉴别诊断 预测模型 

分 类 号：R814.42[医药卫生—影像医学与核医学]