Wu Wenjun, Yao Shengsheng, Liu Daming, Luo Yuan, Sun Yihan, Ruan Ting, Liu Mengyou, Shi Li, Liu Chang, Xiao Mingming, Zhang Qi, Liu Zhengshuai, Ju Xingai, Wang Jiahao, Fei Xiang, Lu Li, Gao Yang, Zhang Ying, Gong Liying, Chen Xuanyu, Zheng Wanli, Niu Xiali, Yang Xiao, Cao Huimei, Chang Shijie, Cui Jianchun, Ma Zuoxin
Liaoning University of Chinese Medicine, Shenyang, China.
Department of Thyroid and Breast Surgery, People's Hospital of China Medical University (Liaoning Provincial People's Hospital), Shenyang, China.
BMC Immunol. 2025 Apr 7;26(1):27. doi: 10.1186/s12865-025-00708-5.
Seronegative Hashimoto's thyroiditis is often underdiagnosed due to the lack of antibody markers. Combining ultrasound radiomics with machine learning offers potential for early detection in patients with normal thyroid function.
Data from 164 patients with single thyroid lesions and normal thyroid function, treated surgically between 2016 and 2024, were retrospectively collected from four hospitals. Radiomics features were extracted from ultrasound images of non-tumorous hypoechoic areas. Pathological lymphocytic infiltration and hypoechoic ratios were evaluated by senior pathologists and ultrasound physicians. A machine learning model, CCH-NET, was developed using a random forest classifier after feature selection with Least Absolute Shrinkage and Selection Operator (LASSO) regression. The model was trained and tested with an 80:20 split and compared to senior ultrasound physicians.
The CCH-NET model achieved a sensitivity of 0.762, specificity of 0.714, and an area under the curve (AUC) of 0.8248, outperforming senior ultrasound physicians (AUC = 0.681). It maintained consistent accuracy across test sets, with F1 scores of 0.778 and 0.720 in Test_1 and Test_2, respectively, and exhibited superior predictive rates.
The CCH-NET model enhances accuracy in detecting early Seronegative Hashimoto's thyroiditis over senior ultrasound physicians.
No. [2023] H013 TRIAL REGISTRATION: Chinese Clinical Trial Registry;CTR2400092179; 12 November 2024.
由于缺乏抗体标志物，血清阴性桥本甲状腺炎常常被漏诊。将超声影像组学与机器学习相结合为甲状腺功能正常的患者实现早期检测提供了可能。
回顾性收集了2016年至2024年间在四家医院接受手术治疗的164例单发甲状腺病变且甲状腺功能正常患者的数据。从非肿瘤性低回声区的超声图像中提取影像组学特征。由资深病理学家和超声医师评估病理淋巴细胞浸润和低回声率。在使用最小绝对收缩和选择算子（LASSO）回归进行特征选择后，采用随机森林分类器开发了一种机器学习模型CCH-NET。该模型以80:20的比例进行训练和测试，并与资深超声医师进行比较。
CCH-NET模型的灵敏度为0.762，特异度为0.714，曲线下面积（AUC）为0.8248，优于资深超声医师（AUC = 0.681）。它在各个测试集上保持了一致的准确性，在测试集1和测试集2中的F1分数分别为0.778和0.720，并且显示出更高的预测率。
CCH-NET模型在检测早期血清阴性桥本甲状腺炎方面比资深超声医师提高了准确性。
编号[2023]H013 试验注册：中国临床试验注册中心；CTR2400092179；2024年11月12日。
