Background: Identifying cervical lymph node metastasis (LNM) in primary thyroid cancer preoperatively using ultrasound is challenging. Therefore, a non-invasive method is needed to assess LNM accurately. Purpose: To address this need, we developed the Primary Thyroid Cancer Lymph Node Metastasis Assessment System (PTC-MAS), a transfer learning-based and B-mode ultrasound images-based automatic assessment system for assessing LNM in primary thyroid cancer. Methods: The system has two parts: YOLO Thyroid Nodule Recognition System (YOLOS) for obtaining regions of interest (ROIs) of nodules, and LMM assessment system for building the LNM assessment system using transfer learning and majority voting with extracted ROIs as input. We retained the relative size features of nodules to improve the system’s performance. Results: We evaluated three transfer learning-based neural networks (DenseNet, ResNet, and GoogLeNet) and majority voting, which had the area under the curves (AUCs) of 0.802, 0.837, 0.823, and 0.858, respectively. Method III preserved relative size features and achieved higher AUCs than Method II, which fixed nodule size. YOLOS achieved high precision and sensitivity on a test set, indicating its potential for ROIs extraction. Conclusions: Our proposed PTC-MAS system effectively assesses primary thyroid cancer LNM based on preserving nodule relative size features. It has potential for guiding treatment modalities and avoiding inaccurate ultrasound results due to tracheal interference.

利用机器学习（machine learning, ML）模型分析桥本甲状腺炎（Hashimoto thyroiditis, HT）合并甲状腺乳头状癌（papillary thyroid carcinoma, PTC）患者的甲状腺肿瘤的二维超声图像提取的影像组学和临床特征，探讨其术前无创识别该类患者颈部淋巴结转移（lymph node metastasis, LNM）的能力。 纳入HT合并PTC患者528例，以病理结果为金标准划分为存在颈部淋巴结转移组和不存在颈部淋巴结转移组，由3名医生独立勾画感兴趣区，提取感兴趣区的影像组学特征，以影像组学特征和影像组学特征结合临床特征2种模式构建随机森林（random forest, RF）、支持向量机（support vector machine, SVM）、LightGBM、K邻近算法（K-nearest neighbor, KNN）和XGBoost模型，在测试集上绘制受试者操作特征（receiver operating characteristic, ROC）曲线评价5种机器学习模型的2种模式的性能，并使用SHapley可加性解释（SHapley Additive exPlanations, SHAP）对模型进行可视化。 5种机器学习模型均具有较好的性能，ROC曲线下面积（area under curve, AUC）为0.798～0.921，其中LightGBM和XGBoost性能最佳，优于其他模型（P<0.05）。影像组学特征结合临床特征构建的机器学习模型优于仅使用影像组学特征构建的模型（P<0.05）。SHAP对性能最佳的模型可视化表明，前后径、上下径、original_shape_VoxelVolume、年龄、wavelet-LHL_firstorder_10Percentile和左右径对LightGBM的影响最显著；上下径、前后径、左右径、original_shape_VoxelVolume、original_firstorder_InterquartileRange和年龄对XGBoost的影响最显著。 基于影像组学和临床特征的机器学习模型能够准确地评估HT合并PTC患者颈部淋巴结状态。在5种机器学习模型中，LightGBM和XGBoost的评估性能最佳。