Enhancing CEUS Value in Papillary Thyroid Carcinoma: A Dual-Model Framework for Patient Selection Based on Conventional Ultrasound Characteristics

Abstract

Objective

To investigate the association between typical contrast-enhanced ultrasound (CEUS) features and conventional ultrasound characteristics in papillary thyroid carcinoma (PTC) and develop predictive models to identify patients who would benefit from CEUS, thereby improving diagnostic efficiency.

Methods

A retrospective analysis was conducted on 155 pathologically confirmed PTC cases (October 2021 to December 2024). Based on CEUS features, patients were divided into a hypoenhanced group (n = 89) and a non-hypoenhanced group (n = 66). Differences in conventional ultrasound features (B-mode, color Doppler and elastography) were compared between the two groups. A logistic regression model and a decision tree model were constructed and validated.

Results

Univariate analysis showed statistically significant differences (p < 0.01) in nodule location, size, morphology, internal echogenicity homogeneity and calcification between the two groups. Both logistic regression and decision tree models identified size and morphology as independent predictors (p < 0.001). The area under the curves (AUCs) for the two models were 0.846 and 0.844, respectively. The decision tree model achieved accuracy rates of 78.6% (training set) and 76.7% (validation set). The fivefold cross-validation results showed that the logistic regression model and the decision tree model achieved mean AUC-receiver operating characteristic values of 0.835 ± 0.054 and 0.719 ± 0.047, and mean accuracy rates of 0.773 ± 0.070 and 0.709 ± 0.039, respectively.

Conclusion

Tumor size and morphology were identified as independent factors associated with characteristic CEUS hypoenhancement in PTC. A 70% probability threshold for hypoenhancement, derived from the node probabilities of our decision tree model, was adopted as a clinically practical criterion for recommending CEUS. Our findings, validated in an independent set, support the following clinical strategy: (i) CEUS is not recommended for suspicious PTC nodules ≤0.5 cm; (ii) CEUS is appropriate as a supplementary diagnostic modality for: (a) suspicious PTC nodules measuring (0.5, 0.9] cm with irregular morphology, and (b) those >0.9 cm irrespective of morphological characteristics.