Related Experiment Video
Updated: Apr 8, 2026
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
Prospective validation of an ultrasound-based thyroid imaging reporting and data system (TI-RADS) on 3980 thyroid
Jing Zhang1, Bo-Ji Liu1, Hui-Xiong Xu2
1Department of Medical Ultrasound, Shanghai Tenth People's Hospital, Tongji University School of Medicine Shanghai, China ; Thyroid Institute, Tongji University School of Medicine Shanghai 200072, China.
Objective:
Ultrasound (US) features of solidity, hypoechogenicity or marked hypoechogenicity, microlobulated or irregular margins, microcalcifications, and taller-than-wide shape are suspicious characteristics for thyroid nodules. An US based Thyroid Imaging Reporting and Data System (TI-RADS) is classified based on the number of aforesaid features. TI-RADS category 3 included nodules without any suspicious features, and categories 4a, 4b, 4c, and 5 included nodules with one, two, three or four, or five suspicious US features. The purpose of the study was to prospectively validate the effectiveness of the TI-RADS.
Methods:
From October 2011 to June 2013, we prospectively categorized 3980 thyroid nodules (3752 benign and 228 malignant lesions) in 2921 patients using TI-RADS classification. TI-RADS categories 2 and 3 were considered as benign whereas TI-RADS categories 4 and 5 as malignant. The sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV) and accuracy were calculated.
Results:
Of the 3980 nodules, 2953 nodules were TI-RADS category 2 (0% malignancy), 466 nodules TI-RADS category 3 (1.3% malignancy), 186 nodules TI-RADS category 4a (4.8% malignancy), 165 nodules TI-RADS category 4b (30.3% malignancy), 188 nodules TI-RADS category 4c (75.5% malignancy), and 22 nodules TI-RADS category 5 (95.5% malignancy). The sensitivity, specificity, PPV, NPV and accuracy were 97%, 90%, 40%, 99%, and 91%, respectively.
Conclusions:
TI-RADS classification had great diagnostic value in diagnosing thyroid nodules. The actual probability of malignancy was in accord with the theory risk of malignancy.
