Interobserver Agreement in Assessing the Sonographic and Elastographic Features of Malignant Thyroid Nodules

OBJECTIVE. The objective of our study was to investigate interobserver agreement for the diagnosis of malignant thyroid nodules using conventional B-mode ultrasound and real-time freehand ultrasound elastography. MATERIALS AND METHODS. Between December 2007 and February 2008, 45 patients (age range, 19-73 years; mean age +/- SD, 45.0 +/- 12.2 years) with 52 thyroid nodules were examined with conventional B-mode ultrasound and real-time freehand ultrasound elastography. All the patients were scheduled to undergo thyroid surgery because a thyroid nodule had been proven malignant on aspiration cytology. Three radiologists independently performed conventional ultrasound and elastography and analyzed the ultrasound images. Using conventional ultrasound, observers recorded the following information about nodular features: composition (solid, cystic, or mixed cystic-solid), echogenicity (hyperechoic, isoechoic, hypoechoic, or markedly hypoechoic), margin (well circumscribed, microlobulated, or irregular), calcification (negative [no calcifications]; microcalcification, macrocalcification, or mixed-type calcifications), and shape (parallel or nonparallel). Observers determined the Ueno classification and area ratio for each nodule using ultrasound elastography. Interobserver agreement was evaluated with Spearman's correlation analysis for all findings except the area ratio, for which Pearson's correlation analysis was used. A p < 0.05 was considered to indicate statistical significance. RESULTS. Statistically significant (p < 0.05) concordance among the three radiologists was found on conventional ultrasound for most features except echogenicity and margin of thyroid nodules. The highest value of concordance on conventional ultrasound was achieved for composition (Spearman's correlation coefficient, 0.70-1.00), followed by shape (0.48-0.79) and calcification (0.47-0.62). The least concordant findings on conventional ultrasound were nodular echogenicity (0.04-0.45) and margin (0.03-0.29). However, there was no statistically significant concordance on elastography for the Ueno classification (Spearman's correlation coefficient, 0.08-0.22; p > 0.05) or the area ratio (Pearson's correlation coefficient, -0.03 to 0.23; p > 0.05). CONCLUSION. Statistically significant concordance among radiologists about most features of malignant thyroid nodules was seen with conventional ultrasound; however, ultrasound elastography did not show reliable interobserver agreement for the diagnosis of malignant thyroid nodules.