갑상선 유두미세암종의 경부 림프절 전이 예측을 위한 초음파 소견의 컴퓨터 분석

Abstract

Purpose To find the ultrasonographic (US) features of papillary thyroid microcarcinoma (PTMC) by visual assessment and computer-assisted analysis (CAA), and to determine useful US features of PTMC can predict cervical lymph node metastasis (LNM).

Materials and Methods. Between June 2004 and May 2009, digital US images of consecutive 144 patients (M: F = 31:113; mean age, 49.7 years) who had pathology-proven PTMC without (n=118, 81.9%) or with LNM (n= 26, 18.1%) were included. Two radiologists retrospectively assessed US findings of PTMC including shape, margin, echogenicity, calcification, location, and the degree of capsular abutment. US features by CAA included texture and calcification features. Texture features were divided into morphologic features (area, perimeter, feret, circulation, curvature (saddle point, elliptic point, parabolic point), orientation, eccentricity, and asymmetry), first order texture features (mean intensity, autocorrelation and histogram (skewness, kurtosis)) and second order texture features (gray-level co-occurrence matrices (GLCM); max, moment, contrast, homogeneity, entropy), and calcification features were divided into the number of calcifications and distribution of calcifications (cluster ratio, area of calcification, calcification distance). All US features were compared between two groups of PTMC without (group 0) or with LNM (group 1) were analyzed by Chi-square test or Fisher’s exact test, and Student t-test.

Results. Among visually assessed US features, markedly hypoechoic echogenicity, the presence of calcification, and isthmic locations showed more frequency in group 1 compared to group 0 (p < .05). Among the US features from CAA, mean intensity, skewness, autocorrelation, and GLCM-max of texture features, and the number of calcifications of the calcification features showed significant difference between the two groups (p < .05).

Conclusion. The computerized US may provide adjunctive value in the prediction of LNM in the cases of PTMC, as well as visual assessment.