Meeting Banner
Abstract #4025

Low-Field Point-of-Care MRI: Automated Estimates of Brain Midline Shift Correlate With Clinical Outcomes in Stroke

Prantik Kundu1, Sadegh M. Salehi1, Bradley A. Cahn2, Mercy H. Mazurek2, Matthew M. Yuen2, Jo Schlemper1, Barbara Gordon-Kundu2, Rafael O'Halloran1, Michal Sofka1, and Kevin N. Sheth2
1Hyperfine Research Inc., Guilford, CT, United States, 2Yale University School of Medicine, New Haven, CT, United States

Low-field (64 mT) point-of-care (POC)-MRI was acquired at the bedside from patients with ischemic and hemorrhagic stroke in the neurointensive care unit at a major academic medical center (n=128). An AI system was trained to quantify brain midline shift (MLS), a standard neuroradiological marker of brain injury from POC-MRI, using anatomical annotations from independent neuroradiologists. A cross-validation experiment showed that AI estimates of MLS from POC-MRI were associated with stroke severity and disability at subsequent discharge. AI estimates of MLS greater than 1.5 mm were positively predictive of poor discharge outcome in the full sample and in ischemic stroke.

This abstract and the presentation materials are available to members only; a login is required.

Join Here