Related Experiment Video
Updated: May 26, 2026
Integrating Augmented Reality Tools in Breast Cancer Related Lymphedema Prognostication and Diagnosis
Published on: February 6, 2020
Interobserver agreement in breast radiological density attribution according to BI-RADS quantitative classification
D Bernardi1, M Pellegrini, S Di Michele
1U.O. Senologia Clinica e Screening Mammografico, Dipartimento di Radiodiagnostica, APSS Trento I, Viale Verona Centro per i Servizi Sanitari, Palazzina C, Piano Terrazza, 38100, Trento, Italy. Daniela.Bernardi@apss.tn.it
Mammography density classification using Breast Imaging Reporting and Data System (BI-RADS) criteria showed high reader reproducibility. However, assigning women to the "dense breast" category varied, suggesting potential for computer-aided density assessment.
Area of Science:
- Radiology
- Medical Imaging
- Oncology
Background:
- Mammographic density is a key breast cancer risk factor.
- Accurate classification of mammographic density is crucial for risk stratification and screening decisions.
Purpose of the Study:
- To evaluate the interobserver agreement among expert mammography readers in classifying breast density using quantitative Breast Imaging Reporting and Data System (BI-RADS) criteria.
- To compare agreement among readers and with a larger panel of experts.
Main Methods:
- Six expert mammography readers assessed 100 mammograms.
- Interobserver agreement was calculated using the kappa statistic on both four-category (D1-D4) and two-category (D1-2 vs. D3-4) bases.
- Agreement was also assessed against a previous panel of 12 readers.
Main Results:
- Substantial to almost perfect agreement was observed between pairs of readers (kappa values ranging from 0.60 to >0.80).
- Agreement between individual readers and the expert panel was also substantial to almost perfect.
- Significant variation (6-15%) was noted among readers when assigning mammograms to the BI-RADS D3-4 (dense breast) category.
Conclusions:
- Visual classification of mammographic density using BI-RADS criteria demonstrates high reproducibility among readers.
- Discrepancies in classifying women as having "dense breasts" (BI-RADS D3-4) persist among readers.
- Further research comparing visual assessment with computer-aided density attribution is recommended, as the latter may offer superior reproducibility.
