基于灰度共生矩阵鉴别小细胞肺癌和非小细胞肺癌

Differentiation of small cell lung cancer and non-small cell lung cancer based on CT gray level co-occurrence matrix

目的:利用CT增强图像灰度共生矩阵进行小细胞肺癌和非小细胞肺癌鉴别的可行性研究。方法:回顾性分析经手术、纤支镜或穿刺活检并经病理证实的小细胞肺癌40例和非小细胞肺癌60例(鳞癌、腺癌各30例)。利用Mazda软件勾画感兴趣区,选取灰度共生矩阵中对比度、相关度、熵、差方差、逆差矩五个纹理特征参数,对服从正态分布的数据采用单因素方差分析,不服从正态分布的数据采用Kruskal-Wallis非参数检验,以P<0.05为差异具有统计学意义。绘制受试者工作特征(ROC)曲线,计算曲线下面积(AUC),比较各参数对小细胞肺癌与非小细胞肺癌的诊断效能。结果:对比度、相关度、差方差及逆差矩在3组肿瘤中的P值均<0.05,差异具有统计学意义,熵P值>0.05,差异无统计学差异。绘制ROC曲线,相关度、逆差矩、相关度与逆差矩的联合预测因子三者具有诊断效能,AUC分别为0.712、0.639、0.758,最佳阈值分别为0.362、0.249、42372.260,对应的敏感度、特异度分别为75.0、61.7;52.5、78.3;72.5、78.3,均有一定的诊断效能,且联合预测因子诊断效能最好。结论:基于常规CT扫描的灰度共生矩阵有助于鉴别小细胞肺癌和非小细胞肺癌,具有一定的临床应用前景。

Objective: To investigate the feasibility of CT gray level co-occurrence matrix to differentiate small cell lung cancer from non-small cell lung cancer. Methods: The CT enhanced images of 40 cases of small cell lung cancer, 60 cases of non-small cell lung cancer(30 cases of squamous cell carcinoma and 30 cases of adenocarcinomas) were analyzed retrospectively. Mazda software was used to delineate the region of interest(ROI). The contrast, correlate, difference variance, inverse difference moment and entropy were extracted in the gray level co-occurrence matrix, which were analyzed by one-way analysis of variance or Kruskal-Wallis nonparametric test. The receiver operating characteristic(ROC) curve was established and area under curve(AUC) was obtained to compare the diagnostic efficacy of each parameter for small cell lung cancer and nonsmall cell lung cancer. Results: The differences of contrast, correlate, difference variance, and inverse difference moment had statistical significance(P<0.05), and the difference of entropy had no statistical significance(P>0.05). The correlate, inverse difference moment and combined predictor of correlate and inverse difference moment had diagnostic efficacy. The AUC were 0.712, 0.639, 0.758, and the optimal thresholds were 0.362, 0.249, and 42 372.260, and sensitivity and specificity were 75.0 and 61.7, 52.5 and 78.3, and 72.5 and 78.3. Conclusion: The CT gray level co-occurrence matrix is helpful to differentiate small cell lung cancer from non-small cell lung cancer, and has certain clinical application prospect.