纹理分析在新型冠状病毒肺炎与普通型肺炎鉴别诊断中的价值  

The value of texture analysis in differential diagnosis of novel coronavirus and common pneumonia

作　　者：杜文壮 蒲如剑 孙霞 付晓琴 王现亮[4] 谭清军[5] 徐文科 DU Wenzhuang;PU Rujian;SUN Xia;FU Xiaoqin;WANG Xianliang;TAN Qingjun;XU Wenke(School of Medical Imaging,Weifang Medical University,Weifang 261053,China;Weifang ADR Monitoring Center;Department of Radiology,Changle Peoples Hospital;Department of Radiology,Weifang Peoples Hospital Department of Radiology,Linqu Peoples Hospital;Department of Radiology,Hanting District in Weifang People's Hospital)

机构地区：[1]潍坊医学院医学影像学院,山东潍坊261053 [2]潍坊市药品不良反应检测中心 [3]昌乐市人民医院放射科 [4]潍坊市人民医院放射科 [5]临朐县人民医院放射科 [6]潍坊市人民医院寒亭院区放射科

出　　处：《潍坊医学院学报》2020年第5期321-324,F0003,共5页Acta Academiae Medicinae Weifang

摘　　要：目的探讨基于胸部CT的纹理分析在新型冠状病毒肺炎(COVID-19)与普通型肺炎鉴别诊断中的价值。方法回顾性分析2020年1月22日~3月5日潍坊地区经临床专家确诊的COVID-19患者32例和同期潍坊市人民医院发热门诊证实为非COVID-19的普通型肺炎患者25例的临床及影像学资料。运用Image J软件分别在各自病变的最大横截面图像上勾画感兴趣区(ROI),测量其直方图参数及灰度共生矩阵参数。比较两组间纹理参数的差异,对有统计学意义的纹理参数值绘制受试者工作特征(ROC)曲线,计算ROC曲线下面积(AUC),预测其鉴别诊断的效能。结果直方图参数中标准差和偏度在两组间比较,差异有显著性(P<0.05),且普通肺炎组均大于COVID-19组。灰度共生矩阵参数中角二矩阵、对比、相关、逆差距及熵在两组间比较,差异均有显著性(P<0.05),且角二矩阵、相关和逆差距COVID-19组值大于普通肺炎组,对比和熵普通肺炎组大于COVID-19组。其中,角二矩阵和熵的AUC最大,AUC为0.978,但熵的特异性(96.67%)高于角二矩阵的特异性(93.33%),熵的鉴别诊断效能更高。结论基于CT图像的纹理分析在COVID-19与普通型肺炎的鉴别诊断中具有一定的价值。Objective To explore the value of CT image texture analysisbased on chest in the differential diagnosis of novel coronavirus and common pneumonia.Methods The clinical and imaging data of 32 patients with COVID-19 confirmed by clinical experts in Weifang from January 22 to March 5,2020 and 25 patients with common pneumonia confirmed as non-COVID-19 in fever clinic of Weifang People’s Hospital at the same time were analyzed retrospectively.The software of Image J was used to delineate the region of interest(ROI) on the largest cross-sectional images of their respective lesions,and the histogram parameters and gray level co-occurrence matrix parameters were measured.Comparing the differences of texture parameters between the two groups,drawing the receiver operating characteristic(ROC) curve for the statistically significant values of texture parameters,calculating the area under curve(AUC) of ROC,and predicting the effectiveness of differential diagnosis.Results The standard deviation and skewness of histogram parameters were significantly different between the two groups(P<0.05),and the common pneumonia group was larger than COVID-19 group.In the parameters of gray level co-occurrence matrix,there are statistical differences between the two groups(P<0.05),and the values of angle matrix,correlation and inverse gap COVID-19 group are larger than those of common pneumonia group,and the contrast and entropy of common pneumonia group are larger than that of COVID-19 group.Among them,the AUC of angle two matrix and entropy is the largest,and the AUC is 0.978,but the specificity of entropy(96.67%) is higher than that of angle two matrix(93.33%),and the differential diagnosis efficiency of entropy is higher.Conclusion Texture analysis based on CT images has certain value in differential diagnosis between novel coronavirus and common pneumonia.

关 键 词：新型冠状病毒 肺炎 纹理分析 计算机成像 鉴别诊断 

分 类 号：R814.42[医药卫生—影像医学与核医学]