Abstract TP262: Low-Field Portable MRI in Intracranial Hemorrhage Patients: Greater White Matter Hyperintensity Volume in Stroke Patients Compared to Controls

Introduction: White matter hyperintensities (WMH) on conventional MRI (cMRI) are linked to stroke risk, mortality, and functional outcome. WMH volume may help guide prognosis and stroke prevention but the cost and logistics of cMRI limit use. Portable MRI (pMRI), a 0.064T mobile scanner, offers a low-cost bedside alternative. Our objective was to determine whether WMH volume can be segmented from pMRI in ICH patients. Methods: In this exploratory retrospective study of pMRI scans from 2019-2023, we included patients with spontaneous ICH who had FLAIR imaging with pMRI during hospitalization. Paired cMRI scans performed during the hospital stay were used for validation. pMRI scans were obtained from age-matched patients, without stroke, scanned due to vascular risk factors. Images were automatically segmented with WMH-SynthSeg. WMH volume was segmented from the contralateral hemisphere to the bleed in ICH patients. In controls, total WMH was halved, yielding average WMH per hemisphere. Analyses were performed using Python 3.9.6. Results: pMRI scans were obtained from 24 ICH patients (mean age 64.4 ± 15.1, 48% male, 68% White, median NIHSS 13), 21 of whom had paired cMRI. They were age-matched within 4 years to 15 controls (mean age 64.4 ± 14.9, 56% male, 76% white). WMH volumes from pMRI correlated with those from cMRI (r=0.776,95% CI [0.518, 0.905], p <0.0001) but showed a median difference (z=52, p=0.026). Bland-Altman analysis of log-transformed WMH volumes showed pMRI measurements averaged 11.6% lower than cMRI (95% CI -57, +81, p=0.86), indicating variability despite correlation. pMRI WMH volume was greater in ICH patients than controls (z=57, p=0.0065; median [IQR] WMH volume 5.33 mL [3.44-7.36 mL] vs. 3.34 mL [2.49-4.44 mL]) with a Hodges–Lehmann median difference of 2.65 mL (95% CI 0.036-3.62 mL). Among ICH patients, 54.2% (95% CI 35.1%–72.1%) had moderate/severe hemispheric WMH (>5mL), compared to 12.0% of controls (95% CI 4.2%–30.0%), corresponding to an odds ratio of 8.67 (95% CI 2.04–36.91, p = 0.0023). The agreement between pMRI and cMRI in identifying moderate/severe WMH was moderate (κ=0.52, CI 0.17-0.84) with a concordance of 76%. Conclusion: Portable MRI produced accurate WMH measurements in ICH patients, who had greater WMH volumes than controls. pMRI is a practical tool for WMH quantification, potentially informing stroke risk and prognostication, though variability and sample size warrant further validation prior to clinical implementation.