Huang Qingshan, Xie Lijun, Huang Liyan, Wei Weili, Li Haiying, Zhuang Yunfang, Liu Xinxiu, Chen Shuqiang, Zhang Sufang
Musculoskeletal Tumor Center, Peking University People's Hospital, Beijing, 100044, People's Republic of China.
Department of Ultrasound, The First Affiliated Hospital of Fujian Medical University, Fuzhou, 350005, People's Republic of China.
Int J Gen Med. 2021 Aug 30;14:5069-5078. doi: 10.2147/IJGM.S331338. eCollection 2021.
High-resolution ultrasound is the first choice for the diagnosis of thyroid nodules, but it is still difficult to distinguish between follicular thyroid carcinoma (FTC) and follicular adenoma (FA). Our research aimed to develop and validate an ultrasonic diagnostic model for differentiating FTC from FA.
This study retrospectively analyzed 196 patients who were diagnosed as FTC (n=83) and FA (n=113). LASSO regression analysis was used to screen clinical and ultrasonic features. Multivariate logistic regression analysis was used to establish the ultrasonic diagnostic model of FTC. Nomogram was used for the visualization of diagnostic models. C-index, ROC, and calibration curves analysis were used to evaluate the accuracy of the diagnostic model. Decision curve analysis (DCA) was used to evaluate the net benefits of the ultrasonic diagnostic model for FTC diagnosis under different threshold probabilities. The bootstrap method was used to verify the ultrasonic diagnostic model.
After Lasso regression analysis, 10 clinical and ultrasonic features were used to construct the ultrasonic diagnostic model of FTC. The C-index and AUC of the model were 0.868 and 0.860, respectively. DCA showed that the ultrasonic model had good clinical application value. The C-index in the validation group was 0.818, which was close to the C-index in the model.
Ultrasonic diagnostic model constructed with 10 clinical and ultrasonic features can better distinguish FTC from FA.
高分辨率超声是诊断甲状腺结节的首选方法，但区分滤泡状甲状腺癌（FTC）和滤泡性腺瘤（FA）仍存在困难。本研究旨在建立并验证一种区分FTC和FA的超声诊断模型。
本研究回顾性分析了196例被诊断为FTC（n = 83）和FA（n = 113）的患者。采用LASSO回归分析筛选临床和超声特征。多元逻辑回归分析用于建立FTC的超声诊断模型。列线图用于诊断模型的可视化。采用C指数、ROC和校准曲线分析评估诊断模型的准确性。决策曲线分析（DCA）用于评估超声诊断模型在不同阈值概率下对FTC诊断的净效益。采用自举法验证超声诊断模型。
经过Lasso回归分析，使用10个临床和超声特征构建了FTC的超声诊断模型。该模型的C指数和AUC分别为0.868和0.860。DCA显示超声模型具有良好的临床应用价值。验证组的C指数为0.818，与模型中的C指数接近。
由10个临床和超声特征构建的超声诊断模型能够更好地区分FTC和FA。
