Evaluation of Interreader Agreement between Optimized Short and Full Breast MRI Protocols for Screening Breast Cancer in Moderate- and High-Risk Women.

Abstract

Introduction: The optimized short breast MRI protocol for breast cancer screening was developed to address some limitations of the complete protocol, such as long examination duration and high cost. This study aimed to evaluate the intraobserver agreement between optimized short and complete protocols in the interpretation of breast MRI images. Specifically, we assessed the acquisition and reading times of the images, as well as the time required for report generation.

Methods: A cross-sectional study was conducted in asymptomatic women with high breast density, at moderate and high risk, who underwent breast MRI in private institutions in Belo Horizonte, MG, Brazil. An optimized short protocol was simulated on a computer by selecting images from the complete examination. The examinations were read independently, with BI-RADS categories assigned. Intraobserver agreement between this shorter protocol and the complete protocol was assessed.

Results: A total of 170 women were included. A 100% BI-RADS agreement rate was observed in categories 1 and 2, an 80.95% agreement rate in category 3, and a 93.75% agreement rate in category 4 between the protocols. The unweighted kappa value was 0.955 (95% confidence interval [CI] 0.916-0.993), while the weighted kappa value was 0.969 (95% CI: 0.943-0.995). Furthermore, the optimized short protocol reduced examination duration by 21 min.

Conclusion: The optimized short breast MRI protocol provides a time-efficient alternative for breast cancer screening in moderate- to high-risk populations with dense breasts while maintaining strong agreement in BI-RADS classification. This protocol reduces costs and enhances patient tolerability, thereby improving accessibility to breast MRI.