Ultrasonic Classification of Multicategory Thyroid Nodules Based on Logistic Regression

This study aims to screen out significant ultrasonic features to establish different predictive models of thyroid nodules based on logistic regression, with different indicators being included and nodular size being differentiated, and then compare them.Ultrasonic features of 1906 thyroid nodules in 1761 patients who had undergone sonography and fine-needle aspiration or surgery in our hospital were retrospectively analyzed. According to nodule size and whether vascular or elastographic indicators being included or not, nodules were divided into 12 groups. By univariate and multivariate analysis, the significant sonographic features to diagnose nodules of each group were screened and compared. The logistic regression models were built, and the cutoff values were calculated. The diagnostic performance of newly established models was validated, and the best model was compared with the American College of Radiology Thyroid Imaging Reporting and Data System.Significant features used to diagnose nodules in all models were hypoechoic, irregular margin, and microcalcification. Predominantly solid was an important indicator to differentiate benign and malignant macronodules. A taller-than-wide shape was a significant indicator of malignant micronodules. Strain elastographic character did show diagnostic value. The area under the curve of logistic regression models for malignant risk prediction were all higher than 0.7, and the best one was model 7, but the diagnostic performance was significantly reduced when models performed bivariate prediction.The most valuable indicators of malignant thyroid nodules are hypoechoic, irregular margin, and microcalcification. New models are suitable for nodules of different sizes and with or without vascular or elastographic features being described.