Fu Ruqian, Deng Shi, Hu Yuting, Luo Peng, Yang Hao, Teng Hua, Zeng Dezhi, Ren Jianli
（ 400010） Department of Ultrasound, The Second Affiliated Hospital of Chongqing Medical University, Chongqing 400010, China.
Sichuan Da Xue Xue Bao Yi Xue Ban. 2024 Jul 20;55(4):1026-1033. doi: 10.12182/20240760605.
To analyze the radiomic and clinical features extracted from 2D ultrasound images of thyroid tumors in patients with Hashimoto's thyroiditis (HT) combined with papillary thyroid carcinoma (PTC) using machine learning (ML) models, and to explore the diagnostic performance of the method in making preoperative noninvasive identification of cervical lymph node metastasis (LNM).
A total of 528 patients with HT combined with PTC were enrolled and divided into two groups based on their pathological results of the presence or absence of LNM. The groups were subsequently designated the With LNM Group and the Without LNM Group. Three ultrasound doctors independently delineated the regions of interest and extracted radiomic features. Two modes, radiomic features and radiomics-clinical features, were used to construct random forest (RF), support vector machine (SVM), LightGBM, K-nearest neighbor (KNN), and XGBoost models. The performance of these five ML models in the two modes was evaluated by the receiver operating characteristic (ROC) curves on the test dataset, and SHapley Additive exPlanations (SHAP) was used for model visualization.
All five ML models showed good performance, with area under the ROC curve (AUC) ranging from 0.798 to 0.921. LightGBM and XGBoost demonstrated the best performance, outperforming the other models (<0.05). The ML models constructed with radiomics-clinical features performed better than those constructed using only radiomic features (<0.05). The SHAP visualization of the best-performing models indicated that the anteroposterior diameter, superoinferior diameter, original_shape_VoxelVolume, age, wavelet-LHL_firstorder_10Percentile, and left-to-right diameter had the most significant effect on the LightGBM model. On the other hand, the superoinferior diameter, anteroposterior diameter, left-to-right diameter, original_shape_VoxelVolume, original_firstorder_InterquartileRange, and age had the most significant effect on the XGBoost model.
ML models based on radiomics and clinical features can accurately evaluate the cervical lymph node status in patients with HT combined with PTC. Among the 5 ML models, LightGBM and XGBoost demonstrate the best evaluation performance.
利用机器学习（ML）模型分析桥本甲状腺炎（HT）合并甲状腺乳头状癌（PTC）患者甲状腺肿瘤二维超声图像提取的影像组学和临床特征，探讨该方法在术前无创识别颈部淋巴结转移（LNM）方面的诊断性能。
共纳入528例HT合并PTC患者，根据有无LNM的病理结果分为两组。随后将两组分别命名为有LNM组和无LNM组。三名超声医生独立勾画感兴趣区域并提取影像组学特征。采用影像组学特征和影像组学-临床特征两种模式构建随机森林（RF）、支持向量机（SVM）、LightGBM、K近邻（KNN）和XGBoost模型。通过测试数据集上的受试者工作特征（ROC）曲线评估这五种ML模型在两种模式下的性能，并使用SHapley加性解释（SHAP）进行模型可视化。
所有五种ML模型均表现出良好性能，ROC曲线下面积（AUC）范围为0.798至0.921。LightGBM和XGBoost表现最佳，优于其他模型（P<0.05）。基于影像组学-临床特征构建的ML模型比仅使用影像组学特征构建的模型表现更好（P<0.05）。表现最佳模型的SHAP可视化表明，前后径、上下径、original_shape_VoxelVolume、年龄、小波-LHL_firstorder_10Percentile和左右径对LightGBM模型影响最为显著。另一方面，上下径、前后径、左右径、original_shape_VoxelVolume、original_firstorder_InterquartileRange和年龄对XGBoost模型影响最为显著。
基于影像组学和临床特征的ML模型能够准确评估HT合并PTC患者的颈部淋巴结状态。在这五种ML模型中，LightGBM和XGBoost表现出最佳评估性能。
