2024 年第 12 期 第 0 卷

临床和超声影像组学及融合模型在桥本甲状腺炎临床进程评价中的应用价值分析

Analysis of the application value of clinical, ultrasound radiomics and fusion model in evaluating the clinical course of Hashimoto′s thyroiditis

作者：何超颖郭佩欣杜慕萱张展东戴志兵

英文作者：He Chaoying Guo Peixin Du Muxuan Zhang Zhandong Dai Zhibing

单位：广州医科大学附属中医医院超声医学科，广州510140

英文单位：Department of Ultrasound the Affiliated Traditional Chinese Medicine Hospital of Guangzhou Medical University Guangzhou 510140 China

关键词：桥本甲状腺炎；超声影像组学特征；临床进程评价；列线图模型

英文关键词：Hashimoto′sthyroiditis;Ultrasoundradiomicsfeatures;Clinicalprocessevaluation;Nomogrammodel

• 摘要：

• 目的 分析临床、超声影像组学及融合模型在桥本甲状腺炎（HT）临床进程评价中的应用价值。方法 本文采用回顾性分析方法。收集2023年2—11月广州医科大学附属中医医院收治的80例HT患者的临床资料，将患者纳入观察组并随机分为训练集（50例）和验证集（30例），另选取同期体检健康者80例纳入对照组。比较2集患者的临床资料和超声影像组学特征。建立影像组学模型。通过单因素和多因素回归方法分析影响患者临床进程的独立危险因素。构建临床-超声影像组学融合模型，采用校正曲线、受试工作者特征（ROC）曲线和临床决策曲线对模型的效能进行评价。结果 2集患者的临床资料和超声影像组学特征差异均无统计学意义（均P＞0.05）。多因素分析结果显示促甲状腺激素、游离甲状腺素、甲状腺过氧化物酶抗体、甲状腺球蛋白抗体、腺体增大、出现弥漫性减低回声、剪切波速度均为影响HT患者临床进程的独立危险因素（均P＜0.05）。融合模型（曲线下面积0.807）的敏感度为73.19%，特异度为72.65%，准确度为74.28%，曲线下面积均高于临床模型和超声影像组学模型（0.718和0.741）。校准曲线和临床决策曲线结果显示，融合模型的拟合效果显著，一致性指数为0.866（95%置信区间：0.702～0.943），其预测性能良好。结论 融合模型在HT临床进程评价中的应用价值高于临床和超声影像组学单独应用，可作为预测HT患者临床进程的临床依据。

• Objective To analyze the application value of clinical, ultrasound radiomics and fusion model in evaluating the clinical course of Hashimoto′s thyroiditis (HT). Methods The retrospective analysis method was used in this article. The clinical data of 80 patients with HT admitted to the Affiliated Traditional Chinese Medicine Hospital of Guangzhou Medical University from February to November 2023 were collected. The patients were included in the observation group and randomly divided into a training set (50 cases) and a validation set (30 cases). Another 80 healthy people in the same period were selected as the control group. The clinical data and ultrasound radiomics features of the two sets of patients were compared. The radiomics model was established. The independent risk factors affecting the clinical course of patients were analyzed by univariate and multivariate regression methods. The clinical ultrasound radiomics fusion model was constructed, and the calibration curve, receiver operating characteristic (ROC) curve and clinical decision curve were used to evaluate the efficacy of the model. Results There was no significant difference in clinical data and ultrasound radiomics features between the two sets of patients (all P>0.05). Multivariate analysis showed that thyroid stimulating hormone, free thyroxine, thyroid peroxidase antibody, thyroglobulin antibody, gland enlargement, diffuse hypoechoic and shear wave velocity were independent risk factors affecting the clinical course of HT patients (all P<0.05). The sensitivity of the fusion model (area under the curve: 0.807) was 73.19%, the specificity was 72.65%, and the accuracy was 74.28%, and the area under the curve was higher than that of the clinical model and ultrasound radiomics model (0.718 and 0.741). The results of calibration curve and clinical decision curve showed that the fitting effect of the fusion model was significant, and the consistency index was 0.866(95% confidence interval: 0.702-0.943), and its prediction performance was good. Conclusion The application value of the fusion model in the evaluation of the clinical process of HT is higher than those of clinical and ultrasound radiomics alone, and it can be used as a clinical basis for predicting the clinical process of HT patients.