Related Experiment Video
Updated: Jun 21, 2025
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
Quantitative ultrasound imaging parameters in patients with cancerous thyroid nodules: development of a diagnostic
1Department of Ultrasound, Xingtai People's Hospital No. 16 Hongxing Street, Xingtai 054500, Hebei, China.
Objective:
This study aimed to develop a diagnostic model utilizing quantitative ultrasound parameters to accurately differentiate benign from malignant thyroid nodules.
Methods:
A retrospective analysis of 194 patients with thyroid nodules, encompassing 65 malignant and 129 benign cases, was performed. Clinical data, ultrasound characteristics, and hemodynamic indicators were compared. Receiver operating characteristic (ROC) curves and logistic regression analysis identified independent diagnostic markers.
Results:
No significant differences in clinical data were observed between the groups (P>0.05). Malignant nodules, however, were more likely to exhibit solid composition, hypoechoicity, irregular shapes, calcifications, central blood flow, and unclear margins (P<0.05). Hemodynamic parameters showed that malignant nodules had lower end-diastolic volume (EDV) but higher peak systolic velocity (PSV), resistive index (RI), and vascularization flow index (VFI) (P<0.001). Independent diagnostic factors identified included calcification, margin definition, RI, and VFI. A risk prediction model was formulated, demonstrating significantly lower scores for benign nodules (P<0.0001), achieving an ROC area of 0.964.
Conclusion:
Color Doppler ultrasound effectively distinguishes malignant from benign thyroid nodules. The diagnostic model emphasizes the importance of calcification, margin clarity, RI, and VFI as critical elements, enhancing the accuracy of thyroid nodule characterization and facilitating informed clinical decisions.
