结直肠癌N分期与淋巴结CT影像表现的多因素回归分析  

Regression analysis on the influencing factors for N staging and CT imaging findings of lymph nodes in colorectal cancer

作　　者：高靳[1] 邹佳瑜[1] 代茂良 周瀚[1] 

机构地区：[1]成都市第三人民医院放射科,四川成都610031

出　　处：《实用放射学杂志》2014年第8期1324-1327,共4页Journal of Practical Radiology

摘　　要：目的：探讨结直肠癌淋巴结转移与其 CT 表现的相关性,建立 CT 预测结直肠癌 N 分期的多因素回归模型。方法经手术病理证实,资料完整的结直肠癌患者53例,记录 CT 图像上区域淋巴结大小、形态、密度、边缘及分布特点,并与病理结果作相关及多因素回归分析。结果预测结直肠癌患者有无淋巴结转移,Logistic 逐步回归选入短径≥8 mm 的淋巴结数、边缘不规则的淋巴结数,预测概率模型敏感性85.7%,特异性91.7%,准确性88.9%；预测是否 N2期,Logistic 逐步回归选入密度不均匀的淋巴结数、边缘模糊的淋巴结数,预测概率模型敏感性60.0%,特异性94.3%,准确性86.7%。结论淋巴结短径≥8 mm、边缘不规则、密度不均匀是结直肠癌 N 分期最重要的指标,组合这些指标并进行权重,可以提高 CT 结直肠癌 N 分期的准确性。Objective To explore the relevance between lymphatic metastasis and CT findings in colorectal cancer,and to establish the logistic regression model to realize CT prediction for N staging in colorectal cancer.Methods Take 53 patients with complete data and colorectal cancer confirmed by pathology,and conduct correlation and regression analysis on pathological results based on the size,shape,density,edge and distribution characteristics of the regional lymph nodes that were expressed on CT images.Results It was predicted that patients with colorectal cancer showed lymph node metastasis,a number of lymph nodes with the minor axis≥ 8 mm and the irregular margin were selected in Logistic stepwise regression,the sensitivity of probabilistic model was predicted to be 85.7 %,specificity was 91.7 % and accuracy was 88.9 %;predict whether it had N2 stage,a number of lymph nodes with the heterogeneous density and the fuzzy margin were selected in Logistic stepwise regression,the sensitivity of probabilistic model was predicted to be 60.0 %,specificity 94.3 % and accuracy 86.7 %.Conclusion Lymph nodes with minor axis ≥ 8 mm,irregular margin,and heterogeneous density were the most important indicators for N staging in colorectal cancer,the combination and weighting of these indicators can improve the accuracy of N staging in colorectal cancer.

关 键 词：结直肠肿瘤 淋巴结 计算机体层成像 回归分析 

分 类 号：R735.3[医药卫生—肿瘤] R814.42[医药卫生—临床医学]