Related Experiment Video
Updated: Feb 28, 2026
Computer-Aided Three-Dimensional Visualization in the Treatment of Locally Advanced Thyroid Cancer
Published on: June 9, 2023
Enhancing Node-RADS for preoperative assessment of cervical lymph node metastases in papillary thyroid carcinoma:
Yanjun Wang1, Jinlu Hou2, Xunjun Chen2
1Nurturing Center of Jiangsu Province for State Laboratory of AI Imaging & Interventional Radiology, Department of Radiology, Zhongda Hospital, Southeast University, Nanjing, China.
Objectives:
To evaluate the diagnostic performance of the Node Reporting and Data System (Node-RADS) and its enhanced versions for central (CLNM) and lateral lymph node metastases (LLNM) in papillary thyroid cancer (PTC).
Materials And Methods:
This retrospective study enrolled 227 patients with histologically confirmed PTC who underwent preoperative thyroid-dedicated contrast-enhanced CT. The original Node-RADS scores were assigned to lymph nodes in cervical levels II-VI, and the results were compared with histopathology. A Hyper‑Enhancement Node‑RADS (HE‑Node‑RADS) model by assigning a hyper‑enhancement feature was further developed. Multivariate logistic regression was used to identify independent predictors of CLNM and construct a Nomogram‑Augmented Node‑RADS (NA‑Node‑RADS). The diagnostic performance of different thresholds of Node-RADS and the modified models was assessed and compared.
Results:
Node-RADS demonstrated moderate to good diagnostic performance for both CLNM (threshold > 1, AUC = 0.741) and LLNM (threshold > 2, AUC = 0.801). Interobserver agreements were excellent (κ = 0.838 for CLNM and κ = 0.827 for LLNM). At threshold > 2, HE-Node-RADS (AUC = 0.883) was significantly superior to Node-RADS for LLNM (p = 0.004). NA-Node-RADS (predicted probability > 47.1%), incorporating age, sex, primary tumor size, and core node features (texture, border, shape), showed superior predictive power for CLNM (AUC = 0.879) compared to Node-RADS (0.795) and HE-Node-RADS (0.807), with good calibration and superior net benefit on decision curve analysis.
Conclusion:
CT‑based Node‑RADS offered robust LNM assessment in PTC. The addition of hyper-enhancement improves Node-RADS performance for LLNM, and the proposed nomogram based on Node-RADS further refines CLNM risk estimation.
Key Points:
Question How about the Node-RADS for lymph node metastasis (LNM) assessment in papillary thyroid cancer (PTC), and can it be improved? Findings Node-RADS demonstrates moderate to good diagnostic efficacy. Hyper‑enhancement Node‑RADS and Nomogram‑Augmented Node‑RADS are significantly superior for lateral and central LNM, respectively. Clinical relevance Node‑RADS is useful for preoperative assessment of LNM in PTC, but thresholds vary by neck compartment. Integrating hyper-enhancement enhances lateral Node-RADS performance, while incorporating clinical factors improves central Node-RADS and may facilitate individualized clinical decision-making.
