Tolga Çukur and colleagues have published “Semi-supervision for clinical contrast-weighted image synthesis from magnetic resonance fingerprinting” in Magnetic Resonance Materials in Physics Biology and Medicine (March 2026).
Magnetic resonance fingerprinting can produce many tissue contrasts from one fast scan, and AI models can turn its data into the standard clinical images that radiologists read. Training such models usually requires long reference scans. The paper shows that semi-supervised training with highly accelerated reference scans, undersampled up to 12 to 16 times, still gives high-quality results, which makes collecting training data across large patient groups much easier.
Authors: Mahmut Yurt, Cagan Alkan, Xiaozhi Cao, Congyu Liao, Zihan Zhou, Tolga Çukur, Ali Syed, John M. Pauly, Shreyas S. Vasanawala, Kawin Setsompop