基于DCE-MRI纹理分析鉴别雌激素受体阳性乳腺癌亚型的初步研究

DCE-MRI Texture Analysis in Differential Diagnosis of Estrogen Receptor (ER) Positive Breast Cancer Molecular Subtypes

目的探讨基于DCE⁃MRI的纹理分析鉴别雌激素受体阳性乳腺癌亚型可行性分析。方法回顾性分析华中科技大学同济医学院武汉中心医院2018年1-10月51例经手术病理证实的雌激素受体阳性乳腺浸润性导管癌患者的术前乳腺DCE⁃MRI图像及临床资料。使用FireVoxel纹理分析软件逐层勾画肿瘤边界,统计得到整个瘤体的纹理分析相关参数:均值、标准偏差、偏度、峰度、不均匀度和熵6个纹理特征以及肿瘤体积大小。比较上述特征在Luminal A型乳腺癌与Luminal B型乳腺癌之间的差异,分析有统计学意义的纹理参数对ER阳性乳腺癌亚型的诊断效能。结果Luminal A型组与Luminal B型组乳腺癌均值、标准偏差、偏度和肿瘤大小差异无统计学意义(P>0.05);峰度、不均匀度和熵可有效区分两种分型乳腺癌,差异有统计学意义(P<0.01),三者诊断恶性肿块的受试者工作特性曲线下面积分别为0.832、0.859和0.891(P<0.01),以熵的鉴别效能最佳,熵≤4.22诊断Luminal B型的敏感度为90.62%,特异度为78.95%。结论基于DCE⁃MRI的纹理分析特征,峰度、不均匀度和熵有助于鉴雌激素受体阳性乳腺癌亚型。熵可成为最佳纹理指标。

Objective To investigate the value of DCE⁃MRI texture analysis in differential diagnosis of es⁃trogen receptor(ER)positive breast cancer molecular subtypes.Methods DCE⁃MRI and clinical data of 51 estro⁃gen receptor(ER)positive breast cancer who underwent examination in Wuhan Center Hospital affiliated to Tongji Medical College of Huazhong University of Science and Technology from January 2018 to October 2018 were retro⁃spectively analyzed.All the patients were assigned into Luminal A and Luminal B breast cancer subtype based on surgical pathology.The FireVoxel software was used to delineated the tumor boundary layer by layer on DCE⁃MRI.The parameters related to texture analysis of the whole tumor were statistically obtained:mean,stddev,skewness,kurtosis,inhomogenity,entropy and volume.The differential efficiency of statistically significant texture parameter for Luminal A and Luminal B breast cancer subtype was analyzed.Results There were no statistical differences in mean,standard deviation,skewness and volume between the two groups(P>0.05).Kurtosis,inhomogeneity and entropy could effectively distinguish between Luminal A and Luminal B breast cancer subtype(P<0.01),and the area under the ROC curve of the three parameters for the Luminal B breast cancer subtype were 0.832、0.859 and 0.891(P<0.01)with entropy as the best.Luminal B breast cancer was diagnosed with entropy≤4.22.The sensitivity and specificity were 90.62%and 78.95%,respectively.Conclusion Among DCE⁃MRI based tex⁃ture analysis features,kurtosis,inhomogeneity and entropyis are helpful for the diagnosis of estrogen receptor(ER)positive breast cancer molecular subtypes.Entropy can be the best texture parameter.