Modified Models for Predicting Malignancy Using Ultrasound Characters Have High Accuracy in Thyroid Nodules With...

Area of Science:

• Oncology
• Radiology
• Medical Diagnostics

Background:

• Current thyroid nodule risk stratification guidelines (EU-TIRADS, 2015ATA) have limitations in accuracy and high-risk patient classification.
• Previous research predominantly focused on papillary thyroid carcinoma (PTC), neglecting follicular thyroid carcinoma (FTC) and medullary thyroid carcinoma (MTC).

Purpose of the Study:

• To evaluate diagnostic accuracy and develop predictive models for thyroid nodule malignancy (PTC, FTC, MTC) using ultrasound features and nodule size.
• To compare the performance of the new model against existing guidelines.

Main Methods:

• Logistic regression analysis applied to data from 1,500 patients in China.
• Inclusion of ten ultrasound characteristics and nodule size for model development.
• Analysis of specific ultrasound features associated with PTC, FTC, and MTC.

Main Results:

• Ultrasound features like taller-than-wide shape and thyroid capsule invasion were significant predictors for PTC.
• Thyroid invasion showed a high odds ratio for MTC, while the halo sign was prominent for FTC.
• The developed logistic model achieved up to 80% accuracy, outperforming TIRADS in categories 4b and 4c.

Conclusions:

• The new logistic model demonstrates improved accuracy in predicting thyroid nodule malignancy, particularly for small nodules.
• This model offers better classification of high-risk patients compared to current guidelines.
• The study provides a more comprehensive approach to thyroid nodule risk assessment, including less common types like FTC and MTC.