| 方如旗,薛晓铃,章国伟,陈霞平,翁淑萍,杨青霞,周作福.建立MRI影像组学列线图术前预测子宫内膜癌浸润深肌层[J].中国医学影像技术,2022,38(4):561~565 |
| 建立MRI影像组学列线图术前预测子宫内膜癌浸润深肌层 |
| MRI radiomics nomogram for preoperative predicting deep myometrial invasion of endometrial cancer |
| 投稿时间：2021-09-07 修订日期：2021-12-07 |
| DOI：10.13929/j.issn.1003-3289.2022.04.020 |
| 中文关键词: 子宫肿瘤 磁共振成像 影像组学 列线图 |
| 英文关键词:uterine neoplasms magnetic resonance imaging radiomics nomograms |
| 基金项目:福建省卫生健康科研人才培养项目（中青年骨干人才培养项目）（2019-ZQN-26）、福建省科技创新联合资金项目（2020Y9146）。 |
|
| 摘要点击次数: 1956 |
| 全文下载次数: 547 |
| 中文摘要: |
| 目的 构建MRI影像组学列线图，评价其预测子宫内膜癌（EC）深肌层浸润（DMI）的价值。方法 分析133例EC患者术前临床及影像学资料，根据术后病理所见肌层浸润情况将其分为DMI组（n=62）和非DMI组（n=71）。获取并筛选其影像组学特征，计算影像组学得分（Radscore），利用多因素logistic回归筛选常规MRI、影像组学及病理学特征，建立DMI列线图预测模型。绘制受试者工作特征（ROC）曲线，计算曲线下面积（AUC），评价并比较常规MRI模型、影像组学模型及列线图预测效能。结果 DMI组与非DMI组患者年龄、肿瘤径线、MR提示的DMI、宫颈间质浸润、病理结果及Radscore差异均有统计学意义（P均<0.05）。多因素logistic回归显示，肿瘤短径、MRI所示DMI及Radscore为预测EC DMI的独立危险因素，所构建列线图模型拟合度良好（P=0.57）。常规MRI模型、影像组学模型及列线图预测EC DMI的AUC分别为0.71、0.78及0.84；列线图的AUC高于常规MRI模型及影像组学模型（P均<0.05）。结论 MRI影像组学列线图可用于术前预测EC DMI。 |
| 英文摘要: |
| Objective To establish nomogram of MRI radiomics, and to observe its value for preoperative predicting deep myometrial invasion (DMI) in patients with endometrial cancer (EC). Methods Clinical and MRI data of 133 EC patients were retrospectively analyzed. According to postoperative pathology, the patients were divided into DMI group (n=62) and non-DMI group (n=71). The radiomics features were extracted and selected to calculate radiomics score (Radscore). The features of conventional MRI, radiomics and pathology were selected with multivariable logistic regression analysis to develop the nomogram for predicting DMI. The receiver operating characteristic (ROC) curves were drawn, and the area under the curve (AUC) were calculated to assess and compare the predictive performances of conventional MRI model, radiomics model and the nomogram. Results Significant differences of age, diameter of tumor, MRI showed DMI, cervical stromal invasion, pathology and Radscore between DMI group and non-DMI group (all P<0.05). Multivariable logistic regression analysis showed the short diameter of tumor, MRI reported DMI and Radscore were the selected independent predictors for DMI. The nomogram showed good fitness (P=0.57). AUC of EC DMI predicted by conventional MRI model, radiomics model and the nomogram was 0.71, 0.78 and 0.84, respectively, of the nomogram was higher than that of conventional MRI and radiomics models (both P<0.05). Conclusion MRI radiomics nomogram could be used to preoperative predict EC DMI. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
