디지털확대유방촬영술에서 미세석회화 분석에 대한 판독자간 일치도

Abstract

PURPOSE: To analyze the interobserver variability of radiologists in their description and final assessment categories of microcalcifications in digital magnification mammographies. MATERIALS AND METHODS: From 2005 to 2006, five radiologists analyzed 66 lesion microcalcifications in 65 patients on digital magnification mammographies using a blind method and including 40 benign and 26 malignant lesions. Each observer evaluated the microcalfication morphology, distribution, and BIRADS(R) category. Using the kappa value, the degree of interobserver agreement was calculated and the rate of malignancy was assessed. RESULTS: The mean kappa value for microcalcification morphology was 0.19, which was considered to be moderate agreement for the microcalcification distribution (k: 0.54). The overall rate of malignancy was 39% for microcalcification morphology and distribution. Among them, amorphous microcalcifications showed the lowest rate of malignancy (17%). The mean kappa value for the final assessment categories of BI-RADS(R) was 0.29 and the mean rate of malignancy was 39%. CONCLUSION: Although there was slight interobserver variability, according to each of the descriptors, the general interobserver agreement in interpretation of microcalcification on digital magnification mammogram was slight to moderate. To improve interobserver agreement for the interpretation of microcalcifications, proper image quality control, standardization of criteria, and proper training of radiologists are needed