Radiology Orals 2025 Abstracts

Purpose: Magnetic resonance Imaging (MRI) of the breast is a powerful diagnostic tool that demonstrates exceptional sensitivity resulting in earlier diagnosis, optimised surgical practice, accurate patient selection for neoadjuvant therapy and tailored treatment regimen.Interpreting MRI images is a complex, multi-parametric activity and therefore accurate diagnosis can be difficult when dealing with a plethora of acquisition types.We present a novel educational and assessment platform BEAM (Breast Education and Assessment in MRI) that aims to optimise breast cancer detection.Methods and Materials: Consultant radiologists in UK and Australia along with scientific experts and industry, came together to build BEAM.The requirements were to produce a cloud-based technology, available to anyone, anywhere with full image interactions across a range of MRI protocols and with the ability to simulate a real life reading experience.Participants needed to be able to navigate through a decision-making tree and ultimately decide whether cases contained a cancer or not and whether relevant findings exist outside the breast.Instant feedback had to be made available as well as personalised AIpowered recommendations to accelerate learning.Results: After 18 months of development, we have BEAM: a rapidly translatable, innovative infrastructure that is based on similar templates we have built for other breast imaging activities such as mammography, digital breast tomosynthesis, ultrasound and contrast-enhanced mammography, and other domains such as lung and prostate imaging.BEAM, through its cloud-based nature enables 24/7 access wherever the clinician is located.Full image interaction and post-processing is available on full-resolution multiparametric images and following diagnosis, clinicians instantly receive detailed assessment of diagnostic skills, tailored interactive feedback and performance benchmarks.BEAM can incorporate AI-powered features to recommend personalised education based on demographics, image interactions, radiomic signatures as well as first-order statistical measures as well as Haralick texture descriptors.The platform is sufficiently versatile to allow anyone to create their own BEAM modules for their own trainees.Conclusion: A clinical, academic and industrial collaboration has resulted in BEAM a universally available education platform to optimise breast MRI interpretation.