摘要
目的:探讨基于动态增强MRI(DCE-MRI)瘤内及瘤周影像组学的列线图在预测早期(cT1-2N0-1M0期)乳腺癌患者ALN负荷中的应用价值.方法:回顾性分析2016年3月-2022年12月经手术病理证实且临床分期为cT1-2N0-1M0期的463例乳腺癌患者的临床病理和MRI影像资料.采用完全随机法以7∶3的比例将患者分为训练集(n=324)和验证集(n=139).使用Radcloud平台提取DCE-MRI图像肿瘤内部和周围3 mm、5 mm和7 mm区域的影像组学特征,通过降维保留纳入模型的最优特征.采用支持向量机分别构建相应的影像组学模型.将单因素分析中P＜0.05的临床特征纳入多因素logistic回归分析,得到与ALN负荷相关的独立危险因素,并以此建立临床模型.最后,基于临床危险因素和瘤内+最佳瘤周影像组学评分(Rad-score)建立列线图模型.采用受试者工作特征(ROC)曲线对模型的预测性能进行评价,并计算出曲线下面积(AUC)、敏感度、特异度和准确度.结果:瘤内模型在训练集和验证集中的AUC分别为0.809、0.762.瘤周特征建模以瘤周5 mm范围结果最佳,在训练集和验证集中的AUC分别为0.745、0.727.多因素logistic回归分析显示肿瘤最大径和MR报告淋巴结状态是预测早期乳腺癌患者ALN高负荷的独立危险因素.进一步结合临床危险因素和Rad-score(瘤内+瘤周5 mm)建立列线图.ROC结果显示该列线图表现出良好的预测性能,在训练集中的AUC、敏感度、特异度和准确度分别为0.875、0.736、0.863、0.825;在验证集中的AUC、敏感度、特异度和准确度分别为0.830、0.692、0.850、0.800.结论:基于DCE-MRI瘤内及瘤周影像组学的列线图能够较好地预测早期乳腺癌患者的ALN负荷.
Abstract
Objective:To explore the application value of dynamic contrast enhanced MRI(DCE-MRI)in predicting ALN burden in early-stage(cT1-2N0-1M0)breast cancer patients based on intra-tumoral and peritumoral radiomics nomogram.Methods:The clinicopathologic and MRI data of 463 with breast cancer patients(cT1-2N0-1M0)confirmed by surgery and pathology from March 2016 to December 2022 were retrospectively analyzed.The patients were divided into a training set(n=324)and a validation set(n=139)in a ratio of 7∶3 using a completely randomized method.The Radcloud platform was used to extract intratumoral and 3mm,5mm and 7mm peritumoral imaging omics fea-tures of DCE-MRI sequence,and the optimal features included in the model were retained through di-mensionality reduction.The support vector machine was used to construct the corresponding radiomics model.Clinical features with P＜0.05 in univariate analysis were incorporated into multivariate logistic regression analysis to obtain independent risk factors correlated with ALN burden,and establish a clinical model based on these factors.Finally,a nomogram model was established based on clinical risk factors and the best intratumoral and peritumoral radiomics score(Rad-score).The predictive per-formance of the model was evaluated with the receiver operating characteristic(ROC)curve,and the area under the curve(AUC),sensitivity,specificity and accuracy were calculated.Results:The AUC of the training set and validation set of the intratumoral model were 0.809 and 0.762,respectively.The AUC of the training set and validation set of peritumoral model,with the optimal range of 5mm around the tumor,were 0.745 and 0.727,respectively.Multivariate logistic regression analysis showed that tumor size and MR reported lymph node status were independent risk factors for ALN burden in pa-tients with early breast cancer.The nomograms were further established by combining clinical risk fac-tors with the Rad-score(intratumoral+5mm peritumoral).The results of the ROC showed that the nomogram had good predictive performance.In the training set,the AUC,sensitivity,specificity and accuracy were 0.875,0.736,0.863,and 0.825,respectively,while in the validation set,the AUC,sensi-tivity,specificity and accuracy were 0.830,0.692,0.850,and 0.800,respectively.Conclusion:The no-mogram based on DCE-MRI intratumoral and peritumoral radiomics could well predict the ALN bur-den in early-stage breast cancer patients.
维普
万方数据