超分辨率超声影像组学列线图术前预测甲状腺乳头状癌侧颈淋巴结转移：开发与验证研究

OBJECTIVE: Super-resolution ultrasound (SRUS) enables the visualization and quantification of nodal microvasculature; however, the existing quantitative parameters are relatively simplistic. This study aimed to assess SRUS-based ultrasomics for diagnosing metastatic lateral cervical lymph nodes (LNs) in papillary thyroid carcinoma (PTC).

METHODS: A total of 171 suspicious lateral cervical LNs from 148 patients with PTC were retrospective included. All LNs underwent B-mode, contrast-enhanced ultrasound (CEUS), and SRUS. Using surgical pathology as the gold standard, radiomics models were constructed by extracting high-throughput features from SRUS images, resulting in three models: the B-mode + CEUS, B-mode + SRUS, and B-mode + CEUS + SRUS models. The optimal radiomics model from the training set was combined with clinical risk factors to construct a nomogram. Discriminatory ability was assessed using the area under the receiver operating characteristic curve (AUC). Calibration curves and decision curve analysis (DCA) were used to evaluate the nomogram's calibration and clinical utility. Finally, the diagnostic performance of the nomogram was compared with that of fine-needle aspiration cytology (FNAC) and thyroglobulin (FNA-Tg) measurements.

RESULTS: Among the 171 LNs, 104 were malignant and 67 were benign. The B-mode + CEUS + SRUS model performed best among three radiomics models (validation set AUC = 0.892). The nomogram demonstrated further improved diagnostic performance (validation set AUC = 0.960), with accuracy significantly superior to that of FNAC (p = 0.027) and an AUC comparable to that of FNA-Tg (p = 0.754). Calibration curves and DCA indicated good calibration and clinical benefit.

CONCLUSION: The SRUS-based radiomics nomogram enables noninvasive, accurate prediction of lateral cervical LN metastasis in PTC, potentially reducing unnecessary biopsies and showing significant clinical value.