The future of breast cancer detection is here (and AI powers it). What if we could detect breast cancer a year earlier?
That’s the tantalizing possibility raised by new research published in Academic Radiology.
Scientists have developed an AI algorithm that shows promise in identifying breast cancer on MRI scans up to 12 months sooner than current methods.
Could this be a game-changer in the fight against this prevalent disease?
As a radiation oncologist who is annually involved in the care of hundreds of patients with breast cancer, the news caught my eye.
The Promise of AI to Improve Cancer Detection
Researchers trained a convolutional neural network AI model using magnetic resonance imaging (MRI) scans from 52,598 breasts.
To refine the model, they used a retrospective dataset of 3,029 MRI scans from 910 high-risk patients (ages 18 to 88; average 52), which included 115 cancers diagnosed within one year of a negative MRI.
The AI model detected cancers one year earlier.
Researchers found that radiologists’ retrospective review of the 10 percent of MRIs the AI deemed highest risk could potentially increase cancer detection by up to nearly one-third (30%).
Study Details
A radiologist could identify visual signs in 83 (72%) biopsy-proven cancer cases.
The AI model correctly identified the anatomical region where the cancer would eventually be detected in 66 (57 percent) of the 115 cases.
My cancer center’s radiologists are remarkably capable of detecting cancer.
The idea that AI can retrospectively find a malignancy from the previous year is exciting.
The breast imaging technology today is remarkable.
AI may allow us to use the device’s output more effectively.
Summary — The Future of Breast Cancer Detection is Here (and AI Powers It).
This novel AI-assisted re-evaluation of “benign” breasts shows promise for improving early breast cancer detection with MRI.
As datasets grow and image quality improves, this approach will be more impactful.
As a radiation oncologist, I’ll end with this: “Cool.”
