Early detection of acute brain injury (ABI) at the bedside is critical in improving survival for patients with extracorporeal membrane oxygenation (ECMO) support. We aimed to examine the safety of ultra–low-field (ULF; 0.064-T) portable magnetic resonance imaging (pMRI) in patients undergoing ECMO and to investigate the ABI frequency and types with ULF-pMRI.

This was a multicenter prospective observational study (SAFE MRI ECMO study [Assessing the Safety and Feasibility of Bedside Portable Low-Field Brain Magnetic Resonance Imaging in Patients on ECMO]; NCT05469139) from 2 tertiary centers (Johns Hopkins, Baltimore, MD and University of Texas–Houston) with specially trained intensive care units. Primary outcomes were safety of ULF-pMRI during ECMO support, defined as completion of ULF-pMRI without significant adverse events.

Of 53 eligible patients, 3 were not scanned because of a large head size that did not fit within the head coil. ULF-pMRI was performed in 50 patients (median age, 58 years; 52% male), with 34 patients (68%) on venoarterial ECMO and 16 patients (32%) on venovenous ECMO. Of 34 patients on venoarterial ECMO, 11 (22%) were centrally cannulated and 23 (46%) were peripherally cannulated. In venovenous ECMO, 9 (18%) had single-lumen cannulation and 7 (14%) had double-lumen cannulation.

Of 50 patients, adverse events occurred in 3 patients (6%), with 2 minor adverse events (ECMO suction event; transient low ECMO flow) and one serious adverse event (intra-aortic balloon pump malfunction attributable to electrocardiographic artifacts). All images demonstrated discernible intracranial pathologies with good quality. ABI was observed in 22 patients (44%). Ischemic stroke (36%) was the most common type of ABI, followed by intracranial hemorrhage (6%) and hypoxic-ischemic brain injury (4%). Of 18 patients (36%) with both ULF-pMRI and head computed tomography within 24 hours, ABI was observed in 9 patients with a total of 10 events (8 ischemic, 2 hemorrhagic events). Of the 8 ischemic events, pMRI observed all 8, and head computed tomography observed only 4 events. For intracranial hemorrhage, pMRI observed only 1 of them, and head computed tomography observed both (2 events).

Our study demonstrates that ULF-pMRI can be performed in patients on ECMO across different ECMO cannulation strategies in specially trained intensive care units. The incidence of ABI was high, seen in 44% of ULF-pMRI studies. ULF-pMRI imaging appears to be more sensitive to ABI, particularly ischemic stroke, compared with head computed tomography.