Abstract
PURPOSE: Our study aimed to diagnose benign or malignant thyroid nodules larger than 4 cm using quantitative diffusion-weighted imaging (DWI) analysis. METHODS: Eighty-two thyroid nodules were investigated retrospectively and divided them into benign (n = 62) and malignant groups (n = 20). We calculated quantitative features DWI and apparent diffusion coefficient (ADC) signal intensity standard deviation (DWI(SD) and ADC(SD)), DWI and ADC signal intensity ratio (DWI(SIR) and ADC(SIR)), mean ADC and minimum ADC value (ADC(mean) and ADC(min)) and ADC value standard deviation (ADC(VSD)). Univariate and multivariate logistic regression were conducted to identify independent predictors, and develop a prediction model. We performed receiver operating characteristic (ROC) analysis to determine the optimal threshold of risk factors, and constructed combined threshold models. Our study calculated diagnostic performance including area under the ROC curve (AUC), accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and unnecessary biopsy rate of all models were calculated and compared them with the American College of Radiology Thyroid Imaging Reporting and Data System (ACR-TIRADS) result. RESULTS: Two independent predictors of malignant nodules were identified by multivariate analysis: DWI(SIR) (P = 0.007) and ADC(min) (P < 0.001). The AUCs for multivariate prediction model, combined DWI(SIR) and ADC(min) thresholds model, combined DWI(SIR) and ADC(SIR) thresholds model and ACR-TIRADS were 0.946 (0.896-0.996), 0.875 (0.759-0.991), 0.777 (0.648-0.907) and 0.722 (0.588-0.857). The combined DWI(SIR) and ADC(min) threshold model had the lowest unnecessary biopsy rate of 0%, compared with 56.3% for ACR-TIRADS. CONCLUSION: Quantitative DWI demonstrated favorable malignant thyroid nodule diagnostic efficacy. The combined DWI(SIR) and ADC(min) thresholds model significantly reduced the unnecessary biopsy rate.