多参数MRI纹理分析预测小病灶乳腺癌腋窝淋巴结转移

Value of multiparametric MRI texture analysis in predicting axillary lymph node metastasis of small-sized breast cancer

目的探讨基于磁共振T_(2)加权像(T_(2)weighted image,T_(2)WI),质子密度加权像(diffusion weighted imaging,DWI)及动态增强磁共振成像(dynamic contrast enhanced-magnetic resonance imaging,DCE-MRI)的多参数磁共振成像(magnetic resonance imaging,MRI)纹理分析在小病灶乳腺浸润性导管癌腋窝淋巴结状态预测中的价值。方法回顾性分析2018年1月至2023年6月于台州市中心医院收治的139例初诊浸润性导管癌患者病历资料。根据术后病理结果分为无腋窝淋巴结转移组85例及腋窝淋巴结转移组54例。所有患者行术前MRI检查,包括T_(2)WI,DWI及DCE-MRI等序列。于各序列肿瘤最大径层面绘制感兴趣区后利用Firevoxel软件进行纹理分析,得出了包括均值、标准差、偏度、峰度和熵在内的5个主要参数。运用单因素分析评估各参数特征值在鉴别腋窝淋巴结状态之间的有效性,将单因素分析有意义的变量采用二元Logistic回归分析以探讨特征值与淋巴结转移状态之间的关联,并绘制受试者操作特征(receiver operating characteristic,ROC)曲线,计算ROC曲线下的面积(area under curve,AUC)。结果纹理参数中DCE-MRI序列所绘感兴趣区得到的熵值及平均值,T_(2)WI的偏度值在两组间差异有统计学意义(P<0.001)。其中DCE-MRI的熵值在单因素分析中AUC值最高为0.719,对所选的参数行多因素分析获得了最佳诊断模型,在鉴别淋巴结转移及无淋巴结转移组中AUC为0.769。结论基于多参数MRI的小病灶乳腺癌纹理分析可以较好地预测术前乳腺癌腋窝淋巴结转移状态。

Objective To investigate the value of multiparametric magnetic resonance imaging(MRI)texture analysis based on T_(2)weighted image(T_(2)WI),diffusion weighted imaging(DWI),and dynamic contrast enhanced-MRI(DCE-MRI)in predicting the axillary lymph node status of small-sized invasive ductal carcinoma(IDC)of the breast.Methods A retrospective analysis was conducted on the medical records of 139 patients with newly diagnosed IDC,who were treated at Taizhou Central Hospital from January 2018 to June 2023.Based on the postoperative pathological results,the patients were divided into two groups:85 cases without axillary lymph node metastasis and 54 cases with axillary lymph node metastasis.All patients underwent preoperative MRI examination,including sequences such as T_(2)WI,DWI,and DCE-MRI.After delineating the region of interest(ROI)on the slice with the largest tumor diameter in each sequence,texture analysis was performed using Firevoxel software,which yielded five major parameters,including mean,standard deviation,skewness,kurtosis,and entropy.Univariate analysis was employed to evaluate the effectiveness of each parameter in distinguishing the axillary lymph node status.Variables that showed significant results in the univariate analysis were then included in binary Logistic regression analysis to explore the relationship between these parameters and lymph node metastasis status.Receiver operating characteristic(ROC)curves were plotted,and the area under the curve(AUC)was calculated.Results Significant differences were observed between the two groups in the entropy and mean values of the ROI delineated on the DCE-MRI sequence,as well as the skewness of the T_(2)WI(P<0.001).Among these texture parameters,the entropy of the DCE-MRI sequence showed the highest AUC value of 0.719 in the univariate analysis.Multivariate analysis of the selected parameters yielded an optimal diagnostic model,with an AUC of 0.769 in differentiating lymph node metastasis from non-metastasis.Conclusion Texture analysis of small-sized breast cancer based on multiparametric MRI can effectively predict the preoperative axillary lymph node status of breast cancer.