At RSNA 2025, Mirai is changing breast cancer screening

If a picture is worth a thousand words, then in artificial intelligence, a picture can be worth millions of data points. AI’s superior ability to analyze images and large volume of imaging-based AI models is why many consider radiology to be among the medical fields most impacted by the rise of AI.

This was no less true at this year’s RSNA Annual Meeting (Nov. 30 – Dec. 4), the world’s largest radiology conference and one of the largest medical meetings overall, where Mirai was prominently featured.

Over the years, Mirai, a deep learning model for 5-year breast cancer risk prediction that was released in 2019, has been the subject of a number of validation studies conducted by clinicians who have been interested in the model’s potential to personalize breast cancer screening.

At this year’s RSNA meeting, validation research on Mirai was shared across nine different sessions in 11 presentations given by clinicians from hospitals across the U.S., marking shift the technology’s adoption that will hopefully only continue to grow.

Below are the 11 abstracts conducting additional validation research on Mirai in different contexts:

S3A-SPBR-4 TRANSFORMER-BASED MULTIMODAL AI RISK MODEL FOR LONG-TERM BREAST CANCER RISK ESTIMATION: PERFORMANCE AND EXTERNAL VALIDATION

S3B-SPBR-7 LONGITUDINAL CHANGE OF MAMMOGRAPHIC AI RISK SCORES OVER REPEATED SCREENING EXAMS

S5-SSBR02-3 ROBUSTNESS OF AN AI SHORT-TERM BREAST CANCER RISK MODEL IN A CLINICAL SETTING WITH DIVERSE MAMMOGRAPHY MODALITIES: PRELIMINARY RESULTS

S5-SSBR02-5 EXTERNAL VALIDATION OF DISCRIMINATION AND CALIBRATION OF AN IMAGE-ONLY AI MODEL FOR 5-YEAR BREAST CANCER RISK PREDICTION ACROSS GEOGRAPHICALLY DIVERSE SCREENING SITES

M2-SPBR-3 IMPROVED 3-YEAR BREAST CANCER RISK PREDICTION BY COMBINING ANATOMICAL NOISE POWER EXPONENT AND MIRAI: A LARGE CLINICAL STUDY

M7-SSBR03-4 IMAGE-ONLY DEEP LEARNING RISK MODEL VS BREAST DENSITY TO PREDICT FALSE NEGATIVE EXAMINATIONS

M7-SSBR03-5 IMPACT OF MAMMOGRAPHIC POSITIONING AND BREAST DENSITY ON AI-BASED SHORT-TERM BREAST CANCER RISK PREDICTION: PRELIMINARY RESULTS

T3-SSBR05-5 PROSPECTIVE FEASIBILITY STUDY OF MIRAI-EXPEDITED EVALUATION FOR HIGH-RISK SCREENING MAMMOGRAMS

W5A-SPBR-6 IDENTIFYING WOMEN AT RISK FOR ADVANCED-STAGE BREAST CANCER USING AN IMAGE-ONLY DEEP LEARNING RISK MODEL

W5A-SPPH-8 ANATOMICAL NOISE POWER EXPONENT AS AN IMAGE-BASED BIOMARKER OF BREAST CANCER RISK: IMPROVING AI MODEL PERFORMANCE – A PILOT STUDY

R2-SPBR-1 EXTERNAL VALIDATION OF BREAST CANCER RISK ASSESSMENT MODEL MIRAI ON DIVERSE SCREENING COHORT