Sun Hye Jeong1, Hyun Sook Hong, Eun Hye Lee. 1. Department of Radiology, Bucheon Hospital, Soonchunhyang University College of Medicine, Bucheon-Si, Republic of Korea.
Abstract
OBJECTIVES: The aim of this study was to evaluate the ultrasonographic features for differentiation of follicular thyroid lesions. METHODS: Ultrasonographic features of surgically confirmed 56 follicular adenoma (FA), 22 follicular carcinoma (FC), and 100 nodular hyperplasia (NH) were evaluated using univariable and multivariable multinomial logistic regression analyses, receiver operating characteristic analyses, and areas under the curve. RESULTS: Tumor diameter, margin, echotexture, cystic changes, calcification, hypoechoic rim, and vascularity were significant on univariable analysis. On multivariable logistic regression analyses, tumor diameter (FA, P = 0.002; odds ratio [OR], 0.75; FC, P = 0.001; OR, 2.02), absence of cystic changes (FA, P = 0.127; OR, 2.21; FC, P ≤ 0.001; OR, 17.74), absence of spongiform appearance (FA, P = 0.234; OR, 0.31; FC, P < 0.001; OR, 1673.46), and peripheral vascularity (FA, P = 0.004; OR, 26.64; FC, P < 0.001; OR, 145060.38) differed significantly among the 3 follicular lesions, with NH as a reference. The areas under the curve for NH, FA, and FC were 0.844, 0.858, and 0.705, respectively, and diagnostic accuracy was 72.6%. CONCLUSIONS: Tumor diameter, cystic changes, spongiform appearance, and peripheral vascularity differed significantly among follicular lesions. The diagnostic capability was moderate.
OBJECTIVES: The aim of this study was to evaluate the ultrasonographic features for differentiation of follicular thyroid lesions. METHODS: Ultrasonographic features of surgically confirmed 56 follicular adenoma (FA), 22 follicular carcinoma (FC), and 100 nodular hyperplasia (NH) were evaluated using univariable and multivariable multinomial logistic regression analyses, receiver operating characteristic analyses, and areas under the curve. RESULTS:Tumor diameter, margin, echotexture, cystic changes, calcification, hypoechoic rim, and vascularity were significant on univariable analysis. On multivariable logistic regression analyses, tumor diameter (FA, P = 0.002; odds ratio [OR], 0.75; FC, P = 0.001; OR, 2.02), absence of cystic changes (FA, P = 0.127; OR, 2.21; FC, P ≤ 0.001; OR, 17.74), absence of spongiform appearance (FA, P = 0.234; OR, 0.31; FC, P < 0.001; OR, 1673.46), and peripheral vascularity (FA, P = 0.004; OR, 26.64; FC, P < 0.001; OR, 145060.38) differed significantly among the 3 follicular lesions, with NH as a reference. The areas under the curve for NH, FA, and FC were 0.844, 0.858, and 0.705, respectively, and diagnostic accuracy was 72.6%. CONCLUSIONS:Tumor diameter, cystic changes, spongiform appearance, and peripheral vascularity differed significantly among follicular lesions. The diagnostic capability was moderate.