Lin Naier, Yu Sihui, Xia Zhipeng, Wang Yifan, Chen Wei, Sha Yan
From the Department of Radiology, Eye & ENT Hospital, Fudan University, Shanghai Medical College, Shanghai, China.
J Comput Assist Tomogr. 2022;46(5):823-829. doi: 10.1097/RCT.0000000000001329. Epub 2022 Jun 3.
The aim of the study was to develop and validate a nomogram model combining radiomic features and clinical characteristics to preoperatively differentiate between low- and high-grade sinonasal squamous cell carcinomas.
A total of 174 patients who underwent diffusion-weighted imaging were included in this study. The patients were allocated to the training and testing cohorts randomly at a ratio of 6:4. The least absolute shrinkage and selection operator regression was applied for feature selection and radiomic signature (radscore) construction. Multivariable logistic regression analysis was applied to identify independent predictors. The performance of the model was evaluated using the area under the receiver operating characteristic curve (AUC), the calibration curve, decision curve analysis, and the clinical impact curve.
The radscore included 9 selected radiomic features. The radscore and clinical stage were independent predictors. The nomogram showed better performance (training cohort: AUC, 0.92; 95% confidence interval, 0.85-0.96; testing cohort: AUC, 0.91; 95% CI, 0.82-0.97) than either the radscore or the clinical stage in both the training and test cohorts ( P < 0.050). The nomogram demonstrated good calibration and clinical usefulness.
The apparent diffusion coefficient-based radiomic nomogram model could be useful in differentiating between low- and high-grade sinonasal squamous cell carcinomas.
本研究旨在开发并验证一种结合影像组学特征和临床特征的列线图模型，以在术前鉴别低级别和高级别鼻窦鳞状细胞癌。
本研究纳入了174例行弥散加权成像的患者。患者按6:4的比例随机分配至训练组和测试组。采用最小绝对收缩和选择算子回归进行特征选择和构建影像组学特征（radscore）。应用多变量逻辑回归分析来识别独立预测因子。使用受试者操作特征曲线下面积（AUC）、校准曲线、决策曲线分析和临床影响曲线来评估模型的性能。
radscore包括9个选定的影像组学特征。radscore和临床分期是独立预测因子。在训练组和测试组中，列线图的表现（训练组：AUC，0.92；95%置信区间，0.85 - 0.96；测试组：AUC，0.91；95%CI，0.82 - 0.97）均优于radscore或临床分期（P < 0.050）。列线图显示出良好的校准和临床实用性。
基于表观扩散系数的影像组学列线图模型可用于鉴别低级别和高级别鼻窦鳞状细胞癌。
