Diagnostic value of thyroid micronodules with high b-value diffusion weighted imaging: Comparative study with high-resolution ultrasound.

2021 
Abstract Purpose This study aims to compare the diagnostic performance of two imaging methods for thyroid nodules ≤1.0 cm and reduce unnecessary overdiagnosis. Methods A retrospective study was conducted on 80 patients with pathologically confirmed solitary thyroid micronodules underwent both high-resolution ultrasound (HRUS) and High b-value (2000 s/mm2) diffusion weighted imaging (DWI). Intra- and interobserver agreement (Intraclass correlation coefficient) was followed by Kruskal-Wallis test to detect whether the quantitative apparent diffusion coefficient (ADC) and thyroid nodule subgroups were related. Cohen's kappa analysis was applied to assess the interobserver consistency of DWI and HRUS characteristics. The receiver operating characteristic curves were adopted for evaluating the diagnostic performance of thyroid malignancy. The sensitivity, specificity, and accuracy of the two imaging methods were compared using the McNemar’s test and Kappa test. Results A total of 80 patients were included, consisting of 43 malignant and 37 benign micronodules. The sensitivity, specificity and accuracy of DWI combined with rADC (ADCmin to ADCn ratio) for the diagnosis of thyroid micronodules were 83.7%, 89.2% and 86.3%, respectively. The area under the curve (AUC) was 0.91 (95% confidence interval [CI]: 0.84–0.97). The sensitivity, specificity and accuracy of HRUS diagnosis were 100%, 62.16% and 82.5%, respectively. Conclusion High b-value DWI is superior to HRUS for evaluating the diagnostic performance of solid thyroid micronodules. DWI and its ADC quantitative analysis could be added to the evaluation of thyroid micronodules to improve the specificity of diagnosis, reduce overdiagnosis and avoid unnecessary biopsies or surgeries.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    32
    References
    0
    Citations
    NaN
    KQI
    []