Elizabeth Lazarus MD, Presenter: Nothing to Disclose

Martha Beretta Mainiero MD, Abstract Co-Author: Nothing to Disclose

Barbara Schepps MD, Abstract Co-Author: Nothing to Disclose

Susan Lyn Koelliker MD, Abstract Co-Author: Nothing to Disclose

Linda S. Livingston MD, Abstract Co-Author: Nothing to Disclose

To measure interobserver variability associated with use of the new BI-RADS ultrasound and mammography lexicon and to assess the positive predictive value of new BI-RADS 4 subcategories.

94 consecutive lesions presenting for image guided biopsy comprised of 59 masses, 32 calcifications, and 3 masses with calcification, were retrospectively evaluated by review of sonographic and mammographic images by 5 radiologists. 52 lesions were evaluated on both mammography and ultrasound, 32 on mammography alone, and 10 on ultrasound alone. Each observer described each lesion using terminology from the BI-RADS 4th edition lexicon and assigned a final BI-RADS category, including the new subcategories of BI-RADS category 4. Inter-observer variability was determined with Cohen’s kappa statistic. Pathologic diagnosis was available for all lesions, and positive predictive value for Category 4 and 5 lesions was determined for all readers combined.

For sonographic descriptors, excellent agreement was obtained for assessment of lesion orientation (kappa value 0.61), shape (kappa value 0.66) and lesion boundary (kappa value 0.69). Moderate agreement was obtained for lesion margin (kappa value 0.40) and posterior acoustic features (kappa value 0.40). Fair agreement was obtained for lesion echo pattern (kappa value 0.29). For mammographic descriptors, moderate agreement was obtained for assessment of mass shape (kappa value 0.48), mass margin (kappa value 0.48), and calcification distribution (kappa value 0.50). Fair agreement was obtained for calcification description (kappa value 0.32). Slight agreement was obtained for mass density (kappa value 0.18). Fair agreement was obtained for final assessment category (kappa value 0.28).Positive predictive values of BI-RADS 4 and 5 assignments were as follows: category 4a: 6/102 (6%), category 4b:17/110 (15%), category 4c: 48/91 (53%) and category 5: 71/78 (91%).

Interobserver agreement using the BI-RADS ultrasound lexicon is moderate to excellent and higher than use of the mammographic lexicon. The final assessment categories 4a, 4b, and 4c are useful in predicting the likelihood of malignancy.

Lazarus, E, Mainiero, M, Schepps, B, Koelliker, S, Livingston, L, BI-RADS Lexicon (4th edition) for Ultrasound and Mammography: Interobserver Variability and Positive Predictive Value.  Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL. http://archive.rsna.org/2004/4407310.html