Observer variability in the application of morphologic and dynamic criteria according to the Breast Imaging Reporting and Data System for MRI / Variação entre observadores na aplicação dos critérios morfológicos e cinéticos propostos pelo BI-RADS® (Breast Imaging Reporting and Data System) para ressonância magnética das mamas

AUTOR(ES)
DATA DE PUBLICAÇÃO

2008

RESUMO

OBJECTIVE. The purpose of this study was to assess the variability among observers with different degree of breast MR expertise in the use of the Breast Imaging Reporting and Data System (BI-RADS®) standardized MR lexicon and to evaluate if the variability in lesion description has any implication over lesion final category and management recommendation before and after the knowledge of clinical and other imaging modalities information. MATERIALS AND METHODS. In 96 women, 109 enhancements, including 36 malignant, were interpreted independently by four radiologists divided in pairs according to their experience in breast MR. Each lesion was described using morphologic and dynamic features from the MR BI-RADS® lexicon. A final category and an overall impression were suggested in two occasions, one before and the other after the information about other breast imaging modalities and clinical examination. The agreement between the two members of each pair was measured concerning the description of the enhancement and both final category and recommendation. Kappa statistics were calculated as measures of agreement beyond chance. In order to determine if the rates of agreement in a group were superior then in the other group, the Wilcoxon statistics were used. RESULTS. The experienced interobserver agreement was substantial (0.64) for BI-RADS® final category after the knowledge of clinical information, moderate for mass shape (0.44), mass enhancement (0.41) and management recommendation after clinical knowledge (0.52). Fair agreement was observed for all descriptors among the observer with less experience and for the other descriptors among the experienced ones. The observers with more experience agreed more between themselves for image descriptors and final assessment and recommendation than did the observers with less experience in breast MR. CONCLUSION. Prior experience in breast MRI decreases variability between observers in the application of the criteria proposed by the BI-RADS for MR. Knowledge of relevant information about clinical and other imaging modalities also increased agreement for final category and management recommendation, especially among the experienced radiologists.

ASSUNTO(S)

magnetic resonance imaging breast imagem por ressonância magnética interobsever agreement medical records variações dependentes do observador mama registros médicos

ACESSO AO ARTIGO

Documentos Relacionados