良恶性肺结节影像学特征及定量参数的鉴别诊断价值

Diagnostic value of imaging features and quantitative parameters of benign and malignant pulmonary nodules

摘要:

目的 分析良恶性肺结节影像学特征及定量参数的鉴别诊断价值，以提高诊断的准确性。方法 回顾性分析本院经手术及穿刺活检病理证实的95例恶性肺结节和44例良性肺结节为研究对象。分析良恶性肺结节CT征象及定量参数的差异，并构Logistic回归模型，绘制受试者工作特征（ROC）曲线并计算出曲线下面积（AUC）鉴别肺结节的良恶性。结果 良恶性肺结节CT征象中支气管征、空泡征、棘突征、胸膜凹陷征差异无统计学意义（P>0.05），分叶征、毛刺征、血管集束征差异有统计学意义（P<0.05）。良恶性肺结节定量参数中病灶的长径、体积占比（体积占比<-300 HU、体积占比-300~50 HU、体积占比>50 HU）、总体积、质量占比（质量占比<-300 HU、质量占比-300~50 HU、质量占比>50 HU）、总质量、最大CT值、最小CT值、平均CT值、标准差、中位数差异无统计学意义（P>0.05），熵、偏度、峰度差异有统计学意义（P<0.05）。Logistic回归分析显示熵、偏度、峰度是鉴别良恶性肺结节的独立预测因子。ROC曲线显示CT定量参数中鉴别诊断价值由高到低为熵的AUC为0.918，阈值为8.28，敏感度和特异度分别为89.47%、88.64%；偏度的AUC为0.812，阈值为-0.95，敏感度和特异度分别为83.16%、75.00%；峰度的AUC为0.881，阈值为7.15，敏感度和特异度分别为88.42%、81.82%。结论 分叶征、毛刺征、血管集束征是恶性肺结节重要的影像学特征；CT定量参数中的熵、偏度、峰度鉴别良恶性肺结节的价值较大。

关键词:良性肺结节;恶性肺结节;CT征象及定量参数;鉴别诊断

Abstract:

Objective The imaging features and quantitative parameters of benign and malignant pulmonary nodules were analyzed in order to improve the accuracy of diagnosis.Methods Totally 95 cases of malignant pulmonary nodules and 44 cases of benign pulmonary nodules confirmed by operation and puncture biopsy were analyzed retrospectively. The difference of CT signs and quantitative parameters of benign and malignant pulmonary nodules was analyzed, and the Logistic regression model was constructed, receiver operating characteristic (ROC) curve was drawn, and area under curve (AUC) was calculated to distinguish benign and malignant pulmonary nodules.Results CT signs of benign and malignant pulmonary nodules showed no significant difference in bronchial sign, vacuole sign, spinous process sign and pleural depression sign (P>0.05). There were significant differences in the lobular sign, burr sign and vascular bunching sign (P<0.05). Quantitative parameters of benign and malignant pulmonary nodules including length diameter, volume ratio (volume ratio <-300 HU, -300–50 HU, >50 HU), total volume, mass ratio (mass ratio <-300 HU, -300–50 HU, >50 HU), total mass, maximum CT value and minimum CT value of the lesions, mean CT value, standard deviation and median had no statistical significance (P>0.05). The differences of entropy, skewness and kurtosis were statistically significant (P<0.05). Logistic regression analysis showed that entropy, skewness and kurtosis were independent predictors of benign and malignant pulmonary nodules. ROC curve showed that the AUC of the differential diagnosis value from high to low was 0.918, the threshold was 8.28, and the sensitivity and specificity were 89.47% and 88.64%, respectively. The AUC of skewness was 0.812, the threshold was -0.95, and the sensitivity and specificity were 83.16% and 75%, respectively. The AUC of kurtosis was 0.881, the threshold was 7.15, and the sensitivity and specificity were 88.42% and 81.82%, respectively.Conclusion Lobular sign, burr sign and vascular bunching sign are important imaging features of malignant pulmonary nodules. The entropy, skewness and kurtosis of CT quantitative parameters are of great value in distinguishing benign and malignant pulmonary nodules.

Key words:benign pulmonary nodules;malignant pulmonary nodules;CT signs and quantitative parameters;differential diagnosis