Classifying Brain Edema with Low-Resolution MRI

• Activity Number: 221 - Contributed Poster Presentations: Section on Statistics in Imaging
• Type: Contributed
• Date/Time: Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Danni Tu* and Dylan Small and Manu S. Goyal and Theodore Satterthwaite and Kelly Clark and Russell Shinohara
• Companies: Univ of Pennsylvania and University of Pennsylvania and Department of Radiology, Washington University School of Medicine and Department of Psychiatry, Perelman School of Medicine, University of Pennsylvania and Penn Statistics in Imaging and Visualization Center, University of Pennsylvania and University of Pennsylvania
• Keywords: Imaging; MRI; biomarkers; low resolution; cerebral malaria

Abstract

For high-resolution, high-quality magnetic resonance images (MRI), state-of-the-art approaches to extract biomarkers from imaging data often work well. However, in low resource settings, the only MRI machines available may have low resolution. In the case of cerebral malaria (a complication of infection by the malaria parasite) in Malawi, the severity of cerebral edema is currently scored based on a low resolution 0.35 tesla MRI. Because manual scoring is time-consuming and there are currently few neuroradiologists in sub-Saharan Africa, it is of interest to automate the scoring. Automation, however, is complicated by the fact that many standard imaging pipelines perform inadequately on low-quality, noisy data. We propose a method to first process this low-quality imaging data in a way that borrows strength from high-quality brain atlases from other studies. We then assess edema using volume- and intensity-based measures. Finally, we develop a classification method to identify severe cases of cerebral edema.