| 韩冬,于楠,张喜荣,吴宏培,任革,吕蕊花,李晨,贺太平.增强CT模型及影像组学模型预测肾透明细胞癌WHO/ISUP分级[J].中国医学影像技术,2021,37(4):582~586 |
| 增强CT模型及影像组学模型预测肾透明细胞癌WHO/ISUP分级 |
| Enhanced CT model and radiomics model for predicting WHO/ISUP grade of clear-cell renal cell carcinoma |
| 投稿时间：2020-03-24 修订日期：2021-03-01 |
| DOI：10.13929/j.issn.1003-3289.2021.04.024 |
| 中文关键词: 癌,肾细胞 肿瘤分级 体层摄影术,X线计算机 影像组学 |
| 英文关键词:carcinoma, renal cell neoplasm grading tomography, X-ray computed radiomics |
| 基金项目:陕西中医药大学学科创新团队建设项目（2019-YS04）。 |
|
| 摘要点击次数: 3216 |
| 全文下载次数: 781 |
| 中文摘要: |
| 目的 比较增强CT模型与影像组学模型预测肾透明细胞癌（ccRCC）WHO/ISUP分级的效能。方法 回顾性分析131例经病理确诊ccRCC患者，按照3 ∶ 2比例分层抽样分为训练集（n=78）和验证集（n=53）。根据2016版肾癌WHO/ISUP病理分级标准，以Ⅰ~Ⅱ级为低级别、Ⅲ~Ⅳ级为高级别ccRCC。训练集55例低级别、23例高级别ccRCC；验证集37例低级别、16例高级别ccRCC。以训练集构建增强CT模型及影像组学模型预测ccRCC级别，于验证集加以验证，比较其诊断效能。结果 增强CT模型在训练集及验证集预测高、低级别ccRCC的曲线下面积（AUC）分别为0.89及0.76，敏感度分别0.83及0.56，特异度分别为0.84及0.87；影像组学模型的AUC分别为0.98及0.85，敏感度分别0.96及0.91，特异度分别为0.75及0.84。训练集中影像组学模型的AUC大于增强CT模型（Z＝2.05，P<0.05），验证集中二者AUC差异无统计学意义（Z＝0.95，P=0.34）。决策曲线分析结果显示高风险概率阈值为0.08~1.00时，影像组学模型净获益高于增强CT模型。结论 影像组学模型预测ccRCC WHO/ISUP分级的效能优于增强CT模型。 |
| 英文摘要: |
| Objective To compare the efficacy of enhanced CT model and radiomics model for predicting WHO/ISUP grade of clear-cell renal cell carcinoma (ccRCC). Methods Data of 131 patients with pathologically confirmed ccRCC were retrospectively analyzed. The patients were divided into training set (n=78) and validation set (n=53) using stratified sampling taken the ratio of 3 ∶ 2. According to the 2016 WHO/ISUP pathological grading standards of renal cancer, grade Ⅰ-Ⅱ ccRCC were classified as low grade whereas Ⅲ-Ⅳ were classified as high grade ccRCC. There were 55 cases of low grade and 23 of high grade ccRCC in the training set, 37 of low grade and 16 of high grade ccRCC in the validation set. Enhanced CT model and radiomics model were constructed in training set to predict ccRCC grade, and were used in validation set. The area under the curve (AUC), the sensitivity and specificity were calculated, and the diagnostic efficiency were compared between two models. Results AUC of enhanced CT model in training set and validation set was 0.89 and 0.76, respectively, and the sensitivity was 0.83 and 0.56, the specificity was 0.84 and 0.87, respectively. AUC of radiomics model in training set and validation set was 0.98 and 0.85, respectively, the sensitivity was 0.96 and 0.91, and the specificity was 0.75 and 0.84, respectively. In training set, AUC of radiomics model was higher than that of the enhanced CT model (Z=2.05, P<0.05), while in validation set, there was no statistical difference of AUC between these two models (Z=0.95, P=0.34). Decision curve analysis showed that the net benefit of radiomics model was superior to that of enhanced CT model within the high risk probability threshold of 0.08-1.00. Conclusion Radiomics model was better than enhanced CT model for predicting WHO/ISUP grade of ccRCC. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
