基于T_(2)WI抑脂瘤内及瘤周水肿影像组学预测三阴性乳腺癌的价值

The Value of Intratumoral and Peritumoral Edema T_(2)WI FS-Based Radiomics in Predicting Triple-Negative Breast Cancer

目的探讨基于T_(2)WI抑脂(FS)瘤内及瘤周水肿的影像组学鉴别三阴性乳腺癌(TNBC)与非三阴性乳腺癌(NTNBC)的价值。方法回顾性分析89例经病理明确诊断为乳腺癌的临床及术前MRI资料,TNBC患者37例,NTNBC患者52例。选取T_(2)WI FS序列逐层手工勾画病灶感兴趣区(ROI),通过联影“uAI Research Portal”平台提取瘤内及瘤周水肿特征。通过最小绝对收缩和选择算法(LASSO)筛选出相关特征,采用五折交叉验证分为训练集及测试集,纳入随机森林分类器,建立基于瘤内、瘤周水肿及瘤内联合瘤周水肿的三个影像组学预测模型。采用受试者工作特征(ROC)曲线评估三个模型的诊断效能。结果TNBC与NTNBC两组患者之间组织学分级及瘤周水肿最大径差异有统计学意义(P<0.05),年龄、绝经状态、肿块最大径、Ki-67表达及腋窝淋巴结状态均无统计学差异(P>0.05)。瘤内联合瘤周水肿建立的影像组学模型具有更好的诊断效能,训练集与测试集曲线下面积(AUC)、灵敏度、特异度及准确率分别为0.974和0.929、96.7%和85.7%、76.2%和90.0%、84.7%和88.2%。结论基于T_(2)WI FS瘤内联合瘤周水肿构建的影像组学模型对鉴别TNBC与NTNBC有较好的价值。

Objective To explore the value of intratumoral and peritumoral edema T_(2)WI FS-based radiomics for identifying triple negative breast cancer(TNBC)and non-triple negative breast cancer(NTNBC).Methods The clinical and MRI before the operation data of 89 cases with breast cancer confirmed by pathology were analyzed retrospectively,including 37 TNBC and 52 NTNBC cases.The T_(2)WI FS was selected to manually outline the region of interest,and the imaging radiomics features of intratumoral and peritumoral edema were extracted in uAI Research Portal platform.The least absolute shrinkage and selection operator(LASSO)was used to screen the optimal radiomics features,and the patients were divided into training and testing cohorts with five-fold cross-validation.Random forest was used as the classifier to establish the intratumoral,peritumoral edema,and intratumoral combined with peritumoral edema radiomics models.The receiver operating curve was used to analyze the predictive performance of each model.Results The differences in the pathological grade and maximum diameter of peritumoral edema were statistically significant between TNBC and NTNBC(P<0.05).Age,menopausal state,maximum diameter of mass,Ki-67 expression and axillary lymph nodes status were not statistically differences(P>0.05).The intratumoral combined with peritumoral edema radiomics model showed the better prediction efficiency,and the area under the curve,diagnostic sensitivity,specificity,and accuracy between training and testing cohorts were 0.974 and 0.929,96.7%and 85.7%,76.2% and 90.0%,84.7% and 88.2%,respectively.Conclusion The radiomics model of intratumoral combined with peritumoral edema based on T2 WI FS has good diagnostic performance in discriminating TNBC and NTNBC.