Magnetic resonance imaging (MRI) is vital in neuroradiology because it enables a noninvasive assessment of brain structure and function. Ultra-high-field 7 Tesla (T) MRI provides higher resolution and better contrast between tissue types compared to routine 3T MRI. This can be helpful, for example, in detecting pathologic changes in the brain. However, 7T MRI scanners are much more expensive and less available in hospitals and research centers. As part of my doctoral research, I will be visiting Harvard Medical School and Boston Children's Hospital (Computational Radiology Laboratory) to develop a cost-effective alternative for the acquisition of 7T MR images. Image super-resolution, which is the process of generating high-resolution images from low-resolution ones, allows to computationally generate high-resolution 7T MRI from low-resolution 3T MRI. I will adapt the deep-learning-based super-resolution method that I have developed for electron microscopy to MRI. The proposed super-resolution method will computationally increase the resolution of 3T MRI to match the resolution of 7T MRI, named the MRI Super-Resolution across Field Strength. Hence, the subsequent MR image analysis and processing tasks can be accomplished more accurately as brain tissue details can be seen in higher resolution and contrast.

2023