| 刘文,刘原庆,李筱菁,胡春洪.MRI影像组学联合临床参数及超声表现预测子宫内膜癌淋巴结转移[J].中国医学影像技术,2026,42(3):387~392 |
| MRI影像组学联合临床参数及超声表现预测子宫内膜癌淋巴结转移 |
| MRI radiomics combined with clinical indexes and ultrasonic features for predicting lymph node metastasis of endometrial carcinoma |
| 投稿时间：2025-11-24 修订日期：2026-03-11 |
| DOI：10.13929/j.issn.1003-3289.2026.03.014 |
| 中文关键词: 癌,子宫内膜样 淋巴转移 磁共振成像 影像组学 |
| 英文关键词:carcinoma,endometrioid lymphatic metastasis magnetic resonance imaging radiomics |
| 基金项目: |
|
| 摘要点击次数: 81 |
| 全文下载次数: 47 |
| 中文摘要: |
| 目的 观察MRI影像组学联合临床参数及超声表现预测子宫内膜癌淋巴结转移(LNM)的价值。方法 回顾性纳入225例子宫内膜癌,包括训练集117例、测试集51例及外部验证集57例;于盆腔MRI中提取并筛选肿瘤影像组学特征、构建MRI影像组学模型;以单及多因素logistic回归分析于临床参数及国际子宫内膜肿瘤分析(IETA)组织定义的超声表现中筛选子宫内膜癌LNM的独立影响因素,分别构建临床及超声模型,并基于上述3个模型构建联合模型。绘制受试者工作特征曲线,计算曲线下面积(AUC),评估联合模型预测子宫内膜癌LNM效能,分别以决策曲线分析(DCA)及校准曲线评估模型净收益及校准度。结果 基于年龄及糖类抗原125(CA125)构建临床模型、基于子宫内膜厚度(EMT)构建超声模型、基于5个最优影像组学特征构建MRI影像组学模型。联合模型预测训练集、测试集和外部验证集子宫内膜癌LNM的AUC分别为0.895、0.870及0.890,与MRI影像组学模型相当(AUC分别为0.852、0.809及0.816,P均＞0.05)并显著高于临床(AUC分别为0.727、0.774及0.731)及超声模型(AUC分别为0.701、0.596及0.723,P均＜0.05);其在0~0.47阈值区间内于训练集的临床净收益高于MRI影像组学模型,且在各集的预测概率更接近真实结果。结论 MRI影像组学联合临床参数及超声表现能较准确地预测子宫内膜癌LNM。 |
| 英文摘要: |
| Objective To explore the value of MRI radiomics combined with clinical indexes and ultrasonic features for predicting lymph node metastasis (LNM) of endometrial carcinoma (EC). Methods Totally 225 EC patients were retrospectively enrolled, including 117 cases in training set, 51 cases in test set and 57 cases in external validation set. Radiomic features of EC were extracted and selected from pelvic MRI to construct MRI radiomics model. Univariate and multivariate logistic regression analyses were performed to screen the independent impact factors of LNM of EC from clinical indexes and ultrasonic features defined by international endometrial tumor analysis (IETA) group, and then clinical and ultrasound models were established, respectively, and a combined model was further constructed based on the above models. Then receiver operating characteristic curve was drawn, and the area under the curve (AUC) was calculated to evaluate the predictive performance of the combined model for EC LNM. Decision curve analysis (DCA) and calibration curves were used to assess the clinical net benefit and calibration of the models, respectively. Results The clinical and ultrasound models were constructed based on age and carbohydrate antigen 125 (CA125), as well as endometrial thickness (EMT), while the MRI radiomics model was established based on 5 optimal radiomics features. AUC of the combined model for predicting EC LNM was 0.895, 0.870 and 0.890 in training set, test set and external validation set, respectively, which were comparable to that of MRI radiomics model (0.852, 0.809 and 0.816, respectively, all P>0.05) and higher than that of clinical model (0.727, 0.774 and 0.731, respectively) and ultrasound model (0.701, 0.596 and 0.723, respectively) (all P＜0.05). Moreover, the combined model achieved higher clinical net benefit within the threshold range of 0—0.47 compared with MRI radiomics model in training set, and its predicted probabilities were closer to the actual outcomes in all sets. Conclusion MRI radiomics combined with clinical index and ultrasonic features could be used to accurately predict EC LNM. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
