通讯作者: 颜小斌, 1317370015@qq.com

DOI：10.12201/bmr.202603.00058

A predictive model for thyroid follicular carcinoma integrating ultrasound and clinical features

• 摘要：目的：探讨基于超声及临床特征的二元多因素Logistic回归分析预测模型在甲状腺滤泡癌（follicular thyroid carcinoma,FTC）中的诊断价值。方法：收集2020年4月～2025年8月于泉州市第一医院行术前超声检查并病理确诊的甲状腺滤泡性肿瘤患者107例，其中甲状腺滤泡性腺瘤组（follicular adenoma, FA）62例与FTC组45例，收集并记录患者临床及超声特征，构建二元多因素Logistic回归分析预测模型，绘制ROC曲线，通过计算曲线下面积，对其诊断效能进行评价。结果：二元Logistic回归分析显示：年龄、甲状腺球蛋白（Thyroglobulin，Tg）、抗甲状腺球蛋白抗体（Thyroglobulin Antibody，TgAb）、成分、回声均匀性 、声晕完整性、声晕厚度、血流丰富程度在两组间存在统计学差异（P<0.05）,上述指标纳入回归分析预测模型，建立的回归分析方程为：Logit (P)=-5.865-0.057×年龄+0.006×Tg+0.012×TgAb+2.181×成分+3.630×回声均匀性+4.309×声晕完整性+4.809×声晕厚度+2.492×血流丰富程度，预测模型ROC（receiver operating characteristic,ROC）曲线下面积为0.936。结论：基于临床特征与超声表现建立的二元多因素 Logistic 回归分析预测模型在甲状腺滤泡性肿瘤良恶性诊断及鉴别诊断中具有重要作用，表现出优异的诊断效能。

  关键词： 甲状腺滤泡癌; 超声; 预测模型; 临床指标

   

  Abstract: Objective: To explore the diagnostic value of binary multivariate Logistic regression analysis prediction model based on ultrasound and clinical features in follicular thyroid carcinoma ( FTC ).Methods: A total of 107 patients with thyroid follicular tumors who underwent preoperative ultrasound examination and pathological diagnosis in the First Hospital of Quanzhou from April 2020 to August 2025 were collected, including 62 cases of follicular adenoma ( FA ) group and 45 cases of FTC group. The clinical and ultrasonic characteristics of the patients were collected and recorded, and the binary multivariate logistic regression analysis prediction model was constructed. The ROC curve was drawn, and the diagnostic efficacy was evaluated by calculating the area under the curve.Results:Binary Logistic regression analysis showed that age, thyroglobulin ( Tg ), anti-thyroglobulin antibody (TgAb), composition, echo uniformity, halo integrity, halo thickness, and blood flow richness were statistically different between the two groups ( P < 0.05 ). The above indicators were included in the regression analysis prediction model. The regression analysis equation was Logit ( P ) = -5.865-0.057 × age + 0.006 × Tg + 0.012 × TgAb + 2.181 × component + 3.630 × echo uniformity + 4.309 × halo integrity + 4.809 × halo thickness + 2.492 × blood flow richness. The area under the ROC curve of the prediction model was 0.936.Conclusion: The binary multivariate Logistic regression analysis prediction model based on clinical features and ultrasound findings showed good diagnostic efficacy in the diagnosis and differential diagnosis of benign and malignant thyroid follicular tumors.

  Key words: Thyroid follicular carcinoma; ultrasound; prediction model; clinical indicators

  提交时间：2026-03-17

  版权声明：作者本人独立拥有该论文的版权，预印本系统仅拥有论文的永久保存权利。任何人未经允许不得重复使用。
• html

• 图表

• 颜小斌, 谢巧捷, 谢思培. 甲状腺滤泡性腺瘤与滤泡癌的临床及超声特征分析. 2026. doi: 10.12201/bmr.202602.00031

  程冬艳, 过新民. 股直肌超声指标联合临床资料评估MHD患者PEW的价值. 2025. doi: 10.12201/bmr.202506.00032

  杨雷, 周梦倩, 汤铜. 术前淋巴结超声特征对甲状腺癌颈侧区淋巴结转移风险的预测. 2024. doi: 10.12201/bmr.202410.00035

  张华, 王霞. 多模态超声与人工智能对甲状腺乳头状癌中央区淋巴结转移诊断中的研究进展. 2025. doi: 10.12201/bmr.202509.00024

  钟立. 血流灌注参数联合高频超声在甲状腺结节性质诊断中的临床价值. 2025. doi: 10.12201/bmr.202504.00041

  王杰, 王至诚, 娄帅, 董建成, 曹新志. 基于深度学习算法Mask R-CNN的甲状腺结节检测模型研究. 2024. doi: 10.12201/bmr.202411.00085

  魏乾坤, 吴芸, 徐新燕, 魏梦珂, 曾智萍, 代宇宁, 帕提曼·艾尼瓦尔, 张晶. 继发性不孕患者IVF-ET临床妊娠的预测模型构建. 2025. doi: 10.12201/bmr.202506.00074

  胡欣. 剖宫产术中寒战风险预测模型及列线图的构建. 2025. doi: 10.12201/bmr.202501.00053

  蔡欣欣, 吴夏阳. 早产儿入院败血症风险预测模型的构建与验证. 2025. doi: 10.12201/bmr.202505.00013

  乔军玲, 骈林萍, 关徐涛, 张印. 关于在超声科临床带教中加强人文教育的思考. 2025. doi: 10.12201/bmr.202503.00026

• 序号	提交日期	编号	操作
  1	2026-03-05	10.12201/bmr.202603.00058V1	下载

• 公开评论  匿名评论  仅发给作者 提交评论