Improving portable ultra-low field magnetic resonance images in patients with acquired brain injury using artificial intelligence

While ultra-low field MRI has lower resolution and signal-to-noise ratio compared to high-field MRI, novel artificial intelligence techniques are being developed that may improve the image quality of the ultra-low field MRI data. In this session, we present results of a study where we used an image-to-image translation deep learning model to improve the quality of ultra-low-field (64mT) MRI scans to generate synthetic high-field (3T) MRI scans in a group of patients with acquired brain injury. This proof-of-concept study offers valuable insights into structural changes in the brain, potentially aiding in lesion identification and in the diagnosis and management of patients with brain injuries.