能谱CT评价胰腺癌淋巴结转移的应用价值  

Application value of spectral CT in evaluating lymph node metastasis of pancreatic ductal adenocarcinoma

作　　者：刘文涛 戴博 张珉[1] 窦新民[1] 吴越[1] 袁军辉[1] LIU Wen-tao;DAI Bo;ZHANG Min;DU Xin-min;WU Yue;YUAN Jun-hui(Henan Tumor Hospital,Zhengzhou 450008,China)

机构地区：[1]河南省肿瘤医院,郑州450008 [2]河南省人民医院放射科

出　　处：《医药论坛杂志》2021年第3期20-22,27,共4页Journal of Medical Forum

基　　金：河南省医学科技攻关计划联合共建项目(LHGJ20190628)。

摘　　要：目的分析能谱CT在检测胰腺癌(pancreatic ductal adenocarcinoma,PDCA)淋巴结转移中的应用价值。方法回顾性分析2019年3月—2020年7月河南省肿瘤医院收治的150例PDCA患者临床资料,所有入选者均经病理检查证实为PDCA,术前均接受能谱CT检查,CT增强扫描使用能谱扫描(GSI)模式,期相为动脉晚期(AP)与门脉期(PP),应用标准算法重建图像,层厚为1.25 mm、5 mm,选取感兴趣区(ROI)以提取病灶部位的纹理特征,使用不同分类统计方法判断患者体内有无淋巴结转移,以判断错误率表示其预测效果。结果本研究150例患者中,5 mm AP图像3种特征选择方法的联合方式(FPM)-非线性判别分析(NDA)方法对淋巴结转移判断错误率最低,判断错误率19.33%(29/150);并且在5mmAp图像中,采用NDA分类统计方法,FPM判断淋巴结转移错误率均低于Fisher系数(Fisher)、互信息(MI)、分类错误概率+平均相关系数(POE+ACC)(P<0.05);采用FPM方法筛选特征,NDA判断淋巴结转移错误率均低于原始数据分析(RDA)、主成分分析(PCA)、线性判别分析(LDA)(P<0.05);不同组合判断错误率对比(P>0.05)。结论 PDCA能谱CT能够作为一种用于预测有无淋巴结转移辅助检查方法。Objective To analyze the application value of spectral CT in detecting lymph node metastasis of PDCA.Methods The clinical data of 150 PDCA patients treated in our hospital from March 2019 to July 2020 were retrospectively analyzed. All the enrolled patients were confirmed to be PDCA by pathological examination, and all of them received energy spectrum CT examination before operation. CT enhanced scanning used the energy spectrum scanning(GSI) mode, with phases of advanced arterial(AP) and portal venous(PP), used standard algorithm to reconstruct the image, the layer thickness was 1.25 mm,5 mm, selected the region of interest(ROI) to extract the texture features of the lesion site, used different classification and statistical methods to determine whether there was lymph node metastasis in the patient’s body, and used the error rate to indicate its prediction effect.Results In the 150 patients in this study, joint mode(FPM)-Nonlinear Discriminant Analysis(NDA) of 5 mmAP image 3 feature selection methods had the lowest judgment error rate for lymph node metastasis, with a judgment error rate of 19.33%(29/150), and in the 5 mm Ap image, used the NDA classified statistical method, FPM judged that the lymph node metastasis error rate was lower than the Fisher coefficient(Fisher), mutual information(MI), classification error probability + average correlation coefficient(POE+ACC)(P<0.05);Used FPM method to select features, NDA judged that the error rate of lymph node metastasis was lower than that of raw data analysis(RDA), principal component analysis(PCA) and linear discriminant analysis(LDA)(P<0.05);Compared judgment error rates of different combinations(P>0.05).Conclusion PDCA can be used as an auxiliary method for predicting the presence or absence of lymph node metastasis.

关 键 词：胰腺癌 能谱CT 淋巴结转移 特征选择 分类统计 

分 类 号：R735.9[医药卫生—肿瘤]