| 基于CT征象的Logistic回归模型预测纯磨玻璃样肺腺癌恶性浸润程度的临床研究 |
| |
| 引用本文： | 邱太春,尹柯,汝晓双,沈晶,陈影影,伍建林. 基于CT征象的Logistic回归模型预测纯磨玻璃样肺腺癌恶性浸润程度的临床研究[J]. 影像诊断与介入放射学, 2019, 0(1): 54-59 |
| |
| 作者姓名： | 邱太春 尹柯 汝晓双 沈晶 陈影影 伍建林 |
| |
| 作者单位： | 大连大学附属中山医院放射科;大连市中心医院放射科 |
| |
| 摘 要： | 目的基于CT征象利用二元Logistic回归建立预测模型来评估表现为纯磨玻璃结节肺腺癌的恶性浸润程度。方法回顾性收集2015年1月～2018年4月大连大学附属中山医院纯磨玻璃结节144例作为模型建立组,收集大连市中心医院纯磨玻璃结节106例作为模型验证组;所有病例均经手术病理证实。模型建立组中,用t检验和卡方检验分析CT征象在浸润前病变和浸润性肺腺癌两组中的差异,把P<0.05的CT征象作为二元Logistic回归的输入变量并建立Logistic回归模型,用模型验证组检验模型诊断效能。结果浸润性肺腺癌组与浸润前病变在大小(11.5+5.1 mm比7.3+1.8 mm)、平均CT值(-569+97 HU比-622+98 HU)差异有统计学意义(P=0.000,0.001),浸润性肺腺癌组多呈不规则形或多角形,边缘多不光滑,病灶内异常空气支气管征,血管Ⅲ型(P=0.001,0.010,0.009,0.000);胸膜凹陷征和空泡/空腔影在鉴别浸润前病灶和浸润性肺腺癌差异没有统计学意义(P=0.338,0.106)。ROC曲线显示大小≥9.0 mm和CT值≥-624 HU是诊断为浸润性肺腺癌的最佳临界值。在模型建立组中,Logistic回归模型的ROC曲线下的面积为0.882(95%可信区间为0.829～0.935),敏感度79.3%,特异度82.7%;在模型验证组中,ROC曲线下面积为0.868(95%可信区间为0.800～0.936),敏感度75.8%,特异度84.1%。结论基于CT征象的Logistic回归模型对于鉴别浸润前病灶、浸润性肺腺癌具有较高的价值。
|
| 关 键 词： | 纯磨玻璃结节 浸润性肺腺癌 LOGISTIC回归 预测模型 |
Logistic regression model based on CT signs to predict invasive pulmonary adenocarcinoma of pure ground-glass nodules |
| |
| QIU Tai-chun,YIN Ke,RU Xiao-shuang,SHEN Jing,CHEN Ying-ying,WU Jian-lin. Logistic regression model based on CT signs to predict invasive pulmonary adenocarcinoma of pure ground-glass nodules[J]. Journal of Diagnostic Imaging & Interventional Radiology, 2019, 0(1): 54-59 |
| |
| Authors: | QIU Tai-chun YIN Ke RU Xiao-shuang SHEN Jing CHEN Ying-ying WU Jian-lin |
| |
| Affiliation: | (Department of Radiology,Zhongshan Hospital Affiliated to Dalian University,Dalian 116001,China) |
| |
| Abstract: | QIU Tai-chun;YIN Ke;RU Xiao-shuang;SHEN Jing;CHEN Ying-ying;WU Jian-lin(Department of Radiology,Zhongshan Hospital Affiliated to Dalian University,Dalian 116001,China) |
| |
| Keywords: | Pure ground glass nodule Invasive pulmonary adenocarcinoma Logistic regression Prediction model |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
