10-fold Deep Learning Accelerated MRI of Joints

• Applicant: Dr. Yannik Leonhardt
• Subject Area: Radiology
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 526174561

Prof. Jan Fritz, Section Chief of Musculoskeletal Radiology at New York University (NYU, host institution), and his research group have successfully developed and clinically tested techniques to significantly accelerate magnetic resonance imaging (MRI) examinations. The limitations of established acceleration techniques such as parallel imaging as well as simultaneous multi-slice imaging and the multiplicative effect of their combination have been evaluated and published in high impact journals such as “Radiology”. With common image reconstruction techniques, acceleration factors above 4 lead to significant losses in image quality due to a decrease in the signal-to-noise ratio, among other factors. Therefore, Artificial Intelligence-based methods are increasingly used for image reconstruction, e.g. for "denoising" or for increasing the resolution. By combining these, potentially much higher factors can be achieved: In preliminary studies, Prof. Fritz's group has already been able to successfully apply a 6- and 8-fold acceleration factor. The goal of the research stay is to develop, optimize and clinically evaluate at least 10-fold Deep Learning-accelerated MRI protocols on joints. We will analyze the effects of a 10-fold acceleration on image quality and how these effects can be reduced by Deep Learning-based reconstruction techniques. After optimization, this protocol will be compared with the clinical standard.

• DFG Programme: WBP Fellowship
• International Connection: USA
• Host: Professor Dr. Jan Fritz