基于CT增强的灰度直方图纹理特征预测胰腺导管腺癌的生存预后  

Pr ediction of survival prognosis of pancreatic ductal adenocarcinoma based on CT enhanced grayscale histogram texture features

作　　者：李梅[1] 胡苗苗 黄菊 Li Mei

机构地区：[1]浙江省湖州市第一人民医院,313000

出　　处：《浙江临床医学》2023年第7期981-983,共3页Zhejiang Clinical Medical Journal

摘　　要：目的基于CT增强的灰度直方图纹理特征与影像学特征建立COX比例风险回归模型预测胰腺导管腺癌(PDAC)患者的生存预后。方法收集经病理证实的PDAC患者104例,记录一般资料和影像学资料,如性别、肿瘤标志物、大小、边缘、形态、淋巴结及远处转移等情况,使用MaZda软件提取门脉期9个灰度直方图纹理特征,用单因素和多因素COX回归模型进行分析。结果单因素COX回归分析显示肿瘤指标、纹理特征Skewness、形状、大小、边界、周围侵犯、远处转移、淋巴转移和门脉期比值为预后预测因素;将上述纳入多因素回归后,得到纹理特征Skewness>1.15、肿瘤标志物升高、肿瘤最大径>5.3 cm、边界不清、淋巴转移和远处转移是PDAC生存预后的独立危险因素;Kurtosis>3.9和门脉期≤0.62是PDAC预后的保护因素。结论PDAC预后差,CT纹理特征和影像学特征结合可以预测PDAC的生存预后。Objective To established a COX proportional hazard regression model to predict the survival and prognosis of pancreatic ductal adenocarcinoma(PDAC)based on CT texture features and imaging findings.Methods 104 cases of PDAC confirmed by pathology were retrospectively analyzed.General clinical and imaging data,such as sex,tumor index,size,edge,shape,lymph node and distant metastasis,were recorded.Nine gray histogram texture features of portal phase were extracted by Mazda software,and analyzed by univariate and multivariate COX regression models.Results Univariate COX regression analysis showed that tumor index,skewness,shape,size,boundary,peripheral invasion,distant metastasis,lymph node metastasis and portal phase ratio were prognostic factors.P<0.1 factors were included in the multivariate regression model,it was found that skewness>1.15,tumor markers increased,maximum diameter of tumor>5.3 cm,unclear boundary,lymph node metastasis and distant metastasis were independent risk factors for survival and prognosis of PDAC.While kurtosis>3.9 and portal phase≤0.62 were protective factors for prognosis of PDAC.Conclusion The prognosis of PDAC is poor,and the combination of CT texture features and imaging features can predict the survival prognosis of PDAC.

关 键 词：胰腺癌 纹理分析 体层摄影术 生存分析 

分 类 号：R735.9[医药卫生—肿瘤]