Evaluation of C-TIRADS combined with artificial inteligence-assisted diagnosis in differential diagnosis of thyroid nodules

ZHENG Meijuan; XUE Ensheng; YU Yue; CHEN Shun; CHEN Cong; FANG Zhenyan

doi:10.19732/j.cnki.2096-6210.2022.05.005

您当前的位置：

首页 >

文章列表页 >

Evaluation of C-TIRADS combined with artificial inteligence-assisted diagnosis in differential diagnosis of thyroid nodules

更新时间：2025-12-15

- Evaluation of C-TIRADS combined with artificial inteligence-assisted diagnosis in differential diagnosis of thyroid nodules
- Vol. 31, Issue 5, Pages: 484-490(2022)
- 作者机构：
- 作者简介：
- 基金信息：
- DOI：10.19732/j.cnki.2096-6210.2022.05.005
  CLC：
- Published Online：28 October 2022，
  
  Published：28 October 2022
- 稿件说明：
移动端阅览
郑梅娟,薛恩生,俞悦,等. C-TIRADS联合人工智能辅诊系统对甲状腺结节良恶性的诊断价值[J]. 肿瘤影像学, 2022, 31(5): 484-490 https://doi.

org/10.19732/j.cnki.2096-6210.2022.05.005
郑梅娟,薛恩生,俞悦,等. C-TIRADS联合人工智能辅诊系统对甲状腺结节良恶性的诊断价值[J]. 肿瘤影像学, 2022, 31(5): 484-490 https://doi. DOI： 10.19732/j.cnki.2096-6210.2022.05.005.

org/10.19732/j.cnki.2096-6210.2022.05.005 DOI：

摘要

目的：

探讨甲状腺结节超声恶性危险分层中国甲状腺影像报告和数据系统（Chinese-Thyroid Imaging Reporting and Data System，C-TIRADS）联合甲状腺结节人工智能（artificial intelligence，AI）辅诊系统对良恶性结节的诊断价值，并分析桥本甲状腺炎（Hashimoto thyroiditis，HT）背景对诊断结果的影响。

方法：

纳入817个甲状腺结节，以组织病理学检查结果为金标准，对比AI辅诊系统、不同年资医师应用C-TIRADS及联合应用的诊断效能；将结节分为HT背景组/非HT背景组，比较结节背景对AI辅诊系统、不同年资医师诊断结果的影响。

结果：

817个甲状腺结节中，恶性462个，良性355个。与高年资医师应用C-TIRADS相比，AI辅诊系统诊断的特异度较高（89.58%

81.69%，

=0.003），准确度两者相当（92.29%

90.58%，

=0.151）；与低年资医师应用C-TIRADS相比，AI辅诊系统诊断的特异度较高（89.58%

56.90%，

=0.002），准确度较高（92.29%

80.05%，

＜0.001）。通过联合AI辅诊系统，高年资医师诊断的特异度提高（81.69%

89.01%，

=0.006），而准确度、灵敏度的差异均无统计学意义（

＞0.05）；低年资医师诊断的特异度（56.90%

86.76%，

＜0.001）及准确度（80.05%

92.66%，

＜0.001）均明显提高，灵敏度的差异无统计学意义（

=0.526）。HT背景下，高年资医师的诊断特异度低于非HT组（69.64%

83.95%，

＜0.05），低年资医师的诊断特异度也低于非HT组（42.86%

59.53%，

＜0.05），但HT背景对AI辅诊系统诊断灵敏度、特异度及准确度均无影响。

结论：

C-TIRADS对甲状腺结节性质的鉴别具有较高的灵敏度；联合AI辅诊系统对高年资医师价值有限，但可提高低年资医师应用C-TIRADS的诊

断特异度及准确度，避免过度诊断。

Abstract

Objective:

To evaluate the application value of Chinese-Thyroid Imaging Reporting and Data System（C-TIRADS）combine artificial intelligence (AI)-assisted diagnosis in the differential diagnosis of benign and malignant thyroid nodules

and to analyze the effect of Hashimoto thyroiditis (HT) on the diagnostic efficacy.

Methods:

This study included 817 thyroid nodules confirmed histopathologically after thyroidectomy and compared the diagnostic efficacy of AI

C-TIRADS and combined application. Grouping the nodules according to the background of HT

then compared the diagnosis results of AI and sonographer with different seniority.

Results:

Of the 817 thyroid nodules

462 were malignant and 355 were benign. Compared with senior sonographer using C-TIRADS

AI system had higher specificity (89.58%

81.69%

=0.003)

comparable accuracy (92.29%

90.58%

=0.151). Compared with junior sonographer using C-TIRADS

AI system had higher specificity (89.58%

56.90%

=0.002)

higher accuracy (92.29%

80.05%

＜0.001). Combined with AI system

the specificity of senior sonographer improved (81.69%

89.01%

=0.006)

while the difference in accuracy and sensitivity were not statistically significant (

＞0.05); and the specificity (56.90%

86.76%

＜0.001) and accuracy (80.05%

92.66%

＜0.001) of junior sonographer improved

while the difference in sensitivity was not statistically significant (

=0.526). With the background of HT

the diagnostic specificity of senior physicians was lower than that of non-HT group (69.64% vs 83.95%

＜0.05)

and the diagnostic specificity of junior physicians was also lower than that of non-HT group (42.86%

59.53%

＜0.05)，but the background of HT had no influence on

the diagnostic sensitivity

specificity and accuracy of AI system.

Conclusion:

C-TIRADS has high sensitivity for the differential diagnosis of thyroid nodules

but low diagnosis specificity. Combined AI system has limited value for senior sonographer

but can improve the diagnostic specificity and accuracy of C-TIRADS for junior sonographer and avoid overdiagnosis.

关键词

Keywords

references

Views

706

下载量

CNKI被引量

Alert me when the article has been cited

提交

Tools

Publicity Resources

Annual progress in ultrasound clinical research on liver diseases in China in 2024

Progress of thyroid ultrasound in diagnosis and treatment in 2022

Multivariate analysis of ultrasonographic characteristics for differentiating benign and malignant BethesdaⅢ category thyroid nodules

Advances in the application of ultrasound artificial intelligence in prostate cancer

Application and significance of multimodal ultrasound for preoperative assessment of metastasis in lymph nodes posterior to the right recurrent laryngeal nerve in papillary carcinoma of the thyroid

Related Author

Hong HAN

Qing LU

Yuli ZHU

Peili FAN

Huixiong XU

ZHANG Ying

ZHANG Yifeng

XU Huixiong

Related Institution

Department of Ultrasound, Zhongshan Hospital of Fudan University

Institute of Ultrasound in Medicine and Engineering of Fudan University

Shanghai Institute of Medical Imaging

Department of Medical Ultrasound, Shanghai Tenth Peoples Hospital, Tongji University

Department of Ultrasound, Zhongshan Hospital, Fudan University

AI问答

⁰