Sunday, December 1 | 12:30 p.m.-1:00 p.m. | BR222-SD-SUA2 | Lakeside, BR Community, Station 2
In this poster presentation, researchers from Brigham and Women's Hospital in Boston will compare interreader agreement between the BI-RADS fourth edition guidelines for breast cancer classification and BI-RADS fifth edition guidelines.Presenter Dr. Leah Portnow will share results from a study that explored the following:
- The effect of BI-RADS fifth edition guidelines on interreader agreement regarding visual mammographic density assessment compared with previously used BI-RADS fourth edition criteria
- The distribution across density categories using the two editions
- The agreement between visual and quantitative volumetric density assessments in the two editions
For the study, seven breast imaging radiologists reviewed 200 screening digital mammograms in two sessions, using the BI-RADS fourth edition in one session and the fifth edition in the second.
The team found that interreader agreement between the two guideline versions was comparable when it came to assessing disease, but statistically significant differences existed between the two versions when it came to assessing density (p < 0.0001): 57% of mammograms were categorized as C (heterogeneously dense) or D (extremely dense) using the fifth edition compared with 46.7% using the fourth edition.
"Breast imaging readers in this study placed significantly more mammograms in the dense categories using BI-RADS 5th edition guidelines, which has implications for patients due to density legislation encouraging enhanced surveillance," the group concluded.












![A normal mammogram confirmed by three-year radiologic follow-up illustrates reader-marked regions of interest (ROIs) during (A) unaided (round 1) and (B) artificial intelligence (AI)–assisted (round 2) reading. Each colored dot represents an ROI for recall by a human reader. Readers could mark more than one ROI per case, represented by multiple dots of the same color. During AI-assisted reading, the AI system displayed three visible prompts: two with suspicion of malignancy scores of 35% (left mediolateral oblique [L MLO] and craniocaudal [L CC]) and one with a suspicion of malignancy score of 10% (right craniocaudal [R CC]), shown as polygonal overlays. Without AI, six of 10 readers (60%) marked a false-positive ROI. With AI assistance, this fell to two of 10 (20%). R MLO = right mediolateral oblique.](https://img.auntminnie.com/mindful/smg/workspaces/default/uploads/2026/07/2026-07-14-radiology-mammogram-ai-auto-bias.H0bYO8QlWs.jpg?auto=format%2Ccompress&fit=crop&h=112&q=70&w=112)






