囊液CT直方图联合常规临床特征在鉴别胰腺囊性肿瘤中的价值  

The Value of Cyst Fluid CT Histogram Analysis Combined with Conventional Clinical Features in Differential Diagnosis of Pancreatic Cystic Tumors

作　　者：施桦 孙双双 刘松[1] 李辉[1] 李琳[2] 施婷婷[1] SHI Hua;SUN Shuangshuang;LIU Song(Department of Radiology,Nanjing Drum Tower Hospital,The Affiliated Hospital of Nanjing University Medical School,Nanjing 210008,P.R.China)

机构地区：[1]南京大学医学院附属鼓楼医院医学影像科,210008 [2]南京大学医学院附属鼓楼医院病理科,210008

出　　处：《临床放射学杂志》2021年第4期736-741,共6页Journal of Clinical Radiology

摘　　要：目的探究囊液CT直方图联合常规临床特征在鉴别胰腺囊性肿瘤中的价值。方法回顾性分析56例胰腺囊性肿瘤患者的临床资料及增强CT图像,其中导管内乳头状粘液瘤(IPMN)21例,浆液性囊腺瘤(SCA)21例,实性假乳头状瘤(SPN)14例。选取静脉期病灶囊性成分面积最大层面,共评估17个CT直方图参数、2个临床特征和6个形态学特征。采用Kruskal Wallis检验和Mann-Whitney U检验分析直方图参数,应用二元Logistic回归分析建立多参数模型,通过受试者工作特征曲线(ROC)分析CT直方图参数和多参数模型对不同胰腺囊性肿瘤的诊断效能。临床和形态学特征采用χ^(2)检验,单组内样本量<5时,采用Fisher精确性检验。结果多个CT直方图参数、临床特征和形态学特征在不同胰腺囊性肿瘤间有显著差异(P<0.05)。多参数模型区分IPMN与non-IPMN、SCA与non-SCA、SPN与non-SPN的曲线下面积分别为0.962、0.849、0.881,准确率分别为92.86%、78.57%、83.93%,较单一的直方图参数鉴别诊断效能更高。结论多个常规临床特征与CT直方图参数在不同胰腺囊性肿瘤间存在显著差异,常规临床特征与直方图参数融合建立多参数模型在整体区分胰腺囊性肿瘤中诊断效能可以进一步提高。Objective To investigate the value of cyst fluid CT histogram analysis combined with conventional clinical features in differential diagnosis of pancreatic cystic tumors. Methods Clinical data and contrast-enhanced CT images of 56 patients with pancreatic cystic tumors were analyzed retrospectively, including 21 cases of intraductal papillary mucinous neoplasm, 21 cases of serous cystadenoma, and 14 cases of solid pseudopapillary neoplasm. The maximum slice of cystic lesion in venous phase was selected, and a total of 17 CT histogram parameters, 2 clinical features and 6 morphological features were evaluated. Kruskal Wallis test and Mann-Whitney U test were used to analyze the histogram parameters, the multivariate modelswere established by binary Logistic regression analysis and the diagnostic performance of CT histogram parameters and multivariate models in differentiating various pancreatic cystic tumors was analyzed using receiver operating characteristic curve analysis. The difference test of clinical and morphological features used χ^(2) test,and Fisher accuracy test was used when samples in a single group were less than 5. Results There were significant differences of multiple CT histogram parameters,clinicalfeatures and morphological features in different pancreatic cystic tumors( P < 0. 05). The multivariate models were used for the differential diagnosis of IPMN vs. non-IPMN,SCA vs. non-SCA,SPN vs. non-SPN. The areas under the curves were 0. 962,0. 849,and 0. 881,and the accuracies were 92. 86%,78. 57%,and 83. 93%,respectively.The diagnostic efficacy of multivariate models was higher than single histogram parameter. Conclusion There were significant differences of multiple conventional clinical features and CT histogram parameters in different pancreatic cystic tumors. The diagnostic efficiency of multi-parameter models established by conventional clinical features and histogram parameters can be further improved in the overall differentiation of pancreatic cystic tumors.

关 键 词：胰腺囊性肿瘤 直方图 体层摄影术 X线计算机 

分 类 号：R735.9[医药卫生—肿瘤] R730.44[医药卫生—临床医学]