BI-RADS lexicon for US and mammography: interobserver variability and positive predictive value

Area of Science:

• Radiology
• Medical Imaging
• Oncology

Background:

• The Breast Imaging Reporting and Data System (BI-RADS) provides standardized terminology for breast imaging.
• Accurate lesion categorization is crucial for patient management and predicting malignancy.
• Evaluating interobserver variability and positive predictive values (PPVs) of BI-RADS categories is essential for refining diagnostic accuracy.

Purpose of the Study:

• To assess interobserver agreement among breast radiologists using the BI-RADS fourth edition terminology for mammographic and sonographic lesions.
• To determine the positive predictive value (PPV) of BI-RADS categories 4a, 4b, and 4c.

Main Methods:

• Retrospective review of 94 lesions in 91 women who underwent image-guided biopsy.
• Five radiologists independently assessed lesions using BI-RADS terminology and assigned BI-RADS categories.
• Interobserver variability was quantified using Cohen kappa statistics; PPVs were calculated for BI-RADS categories.

Main Results:

• Substantial agreement was observed for ultrasonographic descriptors like boundary and shape (kappa > 0.6).
• Moderate agreement was found for mammographic descriptors such as mass shape and margin (kappa ≈ 0.48).
• Positive predictive values increased progressively across BI-RADS categories: 4a (6%), 4b (15%), 4c (53%), and 5 (91%).

Conclusions:

• The BI-RADS fourth edition terminology demonstrates good interobserver agreement, validating its use, particularly the ultrasonographic lexicon.
• BI-RADS subcategories 4a, 4b, and 4c are valuable for predicting the probability of malignancy in breast lesions.