Objective To explore the correlation between the HRCT features and the nature of solitary pulmonary nodules with Logistic regression analysis.Method HRCT scan was performed in 149 cases of solitary pulmonary nodules confirmed by pathology after common CT scan.15 kinds of HRCT features were grouped according to benign or malignant lessions,and χ2 test was undergone.Those features that had significant differences were selected as independent variables,and were analyzed with Logistic regression.The correlation and the correlative degree were analyzed based on the statistical results.Results 8 kinds of HRCT features were selected as independent variables.Logistic regression analysis showed that some HRCT features such as non-/superficial lobulation sign,sharp edge,halo sign and calcification were positively related to benign nodules.The RR value was 15.387,42.162,12.917 and 3.253,respectively.Other HRCT features such as lobulation sign,spicular sign and vascular convergence signs were positively related to malignant nodules.The RR value was 0.067,0.109 and 0.526,respectively.There was no significant difference between benign and malignant nodules in jagged edge.Conclusions The Logistic regression analysis could show the correlation and correlative degree between HRCT features and the nature of lesions,and has important values in differential diagnosis of solitary pulmonary nodules.%目的 应用Logistic回归分析探讨孤立性肺结节病灶HRCT征象与病变性质的相关性及相关程度.方法 收集149例经病理证实的孤立性肺结节行常规和对病灶局部薄层扫描,将初步确定的15项HRCT征象按照病灶良恶性分组进行χ2检验,筛选在统计学上有显著性差异的征象作为自变量做Logistic回归分析,根据统计结果分析HRCT征象与病变性质的相关性及相关程度.结果 选出8项HRCT征象作为自变量做Logistic回归分析,结果显示:无/浅分叶征、边缘锐利、晕征及钙化等征象与良性病变正相关,其RR值分别是15.387,42.162,12.917,3.253;深分叶征、毛刺征及血管集束征等征象与恶性病变呈正相关,其RR值分别是0.067,0.109,0.526;锯齿状边缘在回归分析中无统计学意义.结论 Logistic多元回归分析揭示了肺结节HRCT征象与病灶性质的相关性及相关程度等内在联系,对孤立性肺结节的鉴别诊断有重要价值.
展开▼
