Israeli researchers develop AI method to speed up breast cancer MRI scans
Technion-led study combines deep learning and mathematical modeling to produce dynamic MRI images at up to one frame per second.
A group of researchers from the Technion in Israel and the United States has reported a breakthrough in MRI scanning that could significantly improve breast cancer diagnosis, according to a paper published in Nature Communications.
The researchers developed a new method, called ELITE, that accelerates and enhances MRI scans used in breast cancer imaging, a disease diagnosed in approximately 2.3 million people each year, most of them women. The approach combines artificial intelligence with advanced mathematical modeling to enable dynamic MRI imaging at what the researchers describe as unprecedented speed and accuracy.
The international study brings together expertise in engineering, MRI physics, artificial intelligence and clinical radiology.
Dr. Eddy Solomon of the Technion’s Faculty of Biomedical Engineering, the study’s lead author, said the research focuses on dynamic MRI, a key tool in breast cancer diagnosis. Dynamic MRI is primarily used for screening high-risk populations and is characterized by high sensitivity, with more than 90% accuracy, compared with roughly 50-60% for ultrasound and mammography combined.
However, MRI technology has long faced a fundamental limitation: producing highly detailed images requires relatively long scan times, making it difficult to track the movement of contrast material through tissue in real time. Traditional MRI systems typically generate one image every one to two minutes at best, limiting the ability to capture the rapid dynamics of contrast agents.
Dr. Solomon and his colleagues sought to bridge this gap by combining mathematical modeling that identifies structural and functional patterns in different tissues with a deep neural network (ResNet) trained to remove noise and distortions. The system also reconstructs missing information from undersampled measurements.
The result, according to the researchers, is the ability to generate one image per second.
The improved temporal resolution allows clinicians to track the movement of contrast agents almost continuously. This, the researchers say, could improve the detection of small tumors, help distinguish more accurately between benign and malignant growths, and better characterize tumor biology, including blood flow and vascular permeability.
In a study involving 54 patients, the researchers reported improved tumor visibility compared with existing methods, higher image quality, and strong diagnostic sensitivity.
They also said that shorter scan times could increase the number of patients that can be scanned using a given MRI system, potentially improving access to imaging services.
The findings are presented as a step toward faster and more precise MRI-based cancer diagnostics, though further validation and clinical deployment would be required before broader adoption.
