AI-Driven Real-time Motion and B0-Field Correction

Magnetic resonance imaging (MRI) is one of the most important tools in modern medicine. It allows doctors and researchers to look inside the human body without using radiation. A single MRI examination can last up to 30 minutes or longer, during which the patient inevitably makes small movementssuch as swallowing or slight head motions. Even these minimal movements can cause blurred or unusable images. As a result, scans often need to be repeated, leading to higher costs and unnecessary inconvenience for patients. The goal of this project is to solve one of MRIs biggest challenges: motion-related image errors and magnetic field instabilities during scanning. We combine state-of-the-art MRI physics with artificial intelligence (AI). Our team is developing extremely fast navigators short measurements embedded directly into the main MRI sequence. These navigators can monitor even the smallest head movements and magnetic field changes in real time. Using deep neural networks, the signals are immediately translated into precise information about the head position and the magnetic field state, allowing the MRI system to correct motion while the scan is still running. The new technology will be tested on volunteers at the Medical University of Vienna, at the High-field MR Center, using 3 Tesla and 7 Tesla scanners. These high-performance systems provide exceptionally detailed images of the human brain and enable better understanding of diseases such as multiple sclerosis, brain tumors, and Alzheimers disease. If successful, this project will make MRI examinations faster, more accurate, and more comfortable. It will reduce repeated scans, lower costs, and improve diagnostic quality. Ultimately, the approach will help achieve more precise and reliable diagnoses of neurological disorders and contribute to better patient care.