Development and validation of a novel diagnostic tool for predicting the malignancy probability of thyroid nodules: A retrospective study based on clinical, B-mode, color doppler and elastographic ultrasonographic characteristics.

Background: Clinicians estimate the risk of thyroid nodules and make subsequently decision on the basis of clinical and ultrasonographic findings. Currently, there is no comprehensive diagnostic tool for predicting the malignancy rates of thyroid nodules. Our aim was to develop and validate a novel integrate diagnostic tool for predicting the malignancy probability of thyroid nodules based on clinical, B-mode, Color Doppler and elastographic ultrasonographic characteristics.

Methods: A total of 1016 nodules in 1016 patients who underwent thyroid ultrasonography and surgery from July 2021 to December 2021 were included in this retrospective study. All nodules were confirmed by pathology and randomly classified into the training and validation groups. Clinical, B-mode, Color Doppler and elastographic (CBCE) ultrasonographic characteristics of nodules were recorded. Univariate and multivariate analyses were performed to screen independent predictors associated with thyroid cancer. A multivariate model containing the extracted predictors was constructed and presented in the form of a nomogram. The validation and applicability of the CBCE nomogram was evaluated using the receiver operating characteristic (ROC) curve. Diagnostic performances were calculated to compare the CBCE nomogram with ACR-TIRADS (Thyroid Imaging Reporting Data System by American College of Radiology) and EU-TIRADS (Thyroid Imaging Reporting Data System by European Thyroid Association).

Results: The following factors were included in the CBCE nomogram: patient gender, age, shape, margin, composition and echogenicity, calcification, vascularization distribution, vascularization degree, suspicious lymph node metastases and elastography. The area under the curve (AUC) values were 0.978 and 0.983 for the training and validation groups, respectively. Compared with ACR-TIRADS and EU-TIRADS, the CBCE nomogram showed improved accuracy (0.944) and specificity (0.913) without sacrificing sensitivity (0.963) and showed the highest AUC with an optimal cutoff value of 0.55.

Conclusion: The CBCE nomogram has good and high clinical practicability in predicting the malignancy probability of thyroid nodules.