Comparison of Real-Time and Static Ultrasonography Diagnoses for Detecting Incidental Diffuse Thyroid Disease: A Multicenter Study.

Abstract

This study aimed to compare diagnostic accuracy of real-time and static ultrasonography (US) for differentiating diffuse thyroid disease (DTD) from normal thyroid parenchyma (NTP). At 4 participating institutions, 203 patients underwent real-time thyroid US before thyroid surgery. For static US, the same radiologists retrospectively evaluated US findings on a picture archive and communication system after 4 weeks. In real-time and static US diagnoses, US category included no DTD, indeterminate, suspicious for DTD, and DTD. We investigated the diagnostic accuracy of real-time and static US with a receiver operating characteristic curve analysis using histopathologic results as the reference standard. Histopathologic results exhibited NTP (n = 139), Hashimoto thyroiditis (n = 24), non-Hashimoto lymphocytic thyroiditis (n = 33), and diffuse hyperplasia (n = 7). Of 203 patients, there were significant differences in echogenicity, echotexture, glandular margin, and vascularity of the thyroid gland and US category between NTP and DTD groups in both real-time and static US diagnoses (P < 0.001). The diagnostic indices of real-time and static US were highest when the cutoff criterion was chosen as 1 or more abnormal US features. In addition, US category was the only feature with a significant difference between DTD and NTP groups regardless of the practical experience. The receiver operating characteristic curve analysis showed that real-time US was superior to static US in the diagnostic accuracy; however, there was no significant difference (P = 0.09). In conclusion, real-time and static US can be helpful for detecting incidental DTD by using US classification based on abnormal US features.