Development, standardization, and testing of a lexicon for reporting contrast-enhanced breast magnetic resonance imaging studies

The purpose of this study was to develop, standardize, and test reproducibi lity of a lexicon for reporting contrast-enhanced breast magnetic resonance imaging (MRI) examinations. To standardize breast MRI lesion description a nd reporting, seven radiologists with extensive breast MRI experience devel oped consensus on technical detail, clinical history, and terminology repor ting to describe kinetic and architectural features of lesions detected on contrast-enhanced breast MR Images. This lexicon adapted American College o f Radiology Breast Imaging and Data Reporting System terminology for breast MRI reporting, including recommendations for reporting clinical history, t echnical parameters for breast MRI, descriptions for general breast composi tion, morphologic and kinetic characteristics of mass lesions or regions of abnormal enhancement, and overall Impression and management recommendation s. To test morphology reproducibility, seven radiologists assessed morpholo gy characteristics of 85 contrast-enhanced breast MRI studies. Data from ea ch independent reader were used to compute weighted and unweighted kappa (k appa) statistics for interobserver agreement among readers. The MR lexicon differentiates two lesion types, mass and non-mass-like enhancement based o n morphology and geographical distribution, with descriptors of shape, marg in, and internal enhancement. Lexicon testing showed substantial agreement for breast density (kappa = 0.63) and moderate agreement for lesion type (k appa = 0.57), mass margins (kappa = 0.55), and mass shape (kappa = 0.42). A greement was fair for internal enhancement characteristics. Unweighted kapp a statistics showed highest agreement for the terms dense in the breast com position category, mass in lesion type, spiculated and smooth In mass margi ns, irregular In mass shape, and both dark septations and rim enhancement f or internal enhancement characteristics within a mass. The newly developed breast MR lexicon demonstrated moderate interobserver agreement. While brea st density and lesion type appear reproducible, other terms require further refinement and testing to lead to a uniform standard language and reportin g system for breast MRI. (C) 2001 Wiley-Liss, Inc.