Prognostication with Low-field Portable MRI: White Matter Hyperintensity Volume Predicts Early Outcomes after Acute Ischemic Stroke
Annabelle Shanks1, Hailey Brigger1, Steph Maynez1, Ian Johnson1, Alison Champagne1, Gordon Sze2, Sam Payabvash2, Jua Iglesias Gonzalez3, Matthew Rosen3, Annabel Sorby-Adams4, W. Kimberly4, Adam De Havenon1, Kevin Sheth1
1Department of Neurology, 2Department of Radiology, Yale New Haven Hospital and Yale School of Medicine, 3Athinoula A. Martinos Center for Biomedical Imaging, 4Department of Neurology, Massachusetts General Hospital and Harvard Medical School
Objective:
To assess agreement between low-field portable MRI (pMRI) and conventional MRI (cMRI) white matter hyperintensity volumes (WMHv) and determine whether portable WMHv predicts outcome after acute ischemic stroke (AIS).
Background:
WMHv ascertained from cMRI is associated with higher AIS risk, greater severity, and worse outcome. Given cMRI’s logistical and financial constraints, pMRI may offer a practical, scalable alternative to cMRI.
Design/Methods:
We retrospectively analyzed pMRI scans with FLAIR sequences from AIS admissions (2019-2024) and used cMRI scans from the same hospitalization for validation. WMHv was quantified automatically using WMH-SynthSeg. Ordinal logistic models adjusted for age, sex, admission NIHSS, and vascular risk factors predicted discharge modified Rankin Scale (mRS) and disposition. A binary logistic regression with the same adjustments predicted 3-month mRS (poor outcome mRS>1). Analyses were performed using Python 3.9.6.
Results:
pMRI scans were obtained from 155 patients (mean age 64.4 ± 15, 59.4% male, 65.8% white, median NIHSS 9), 131 of whom received a cMRI (avg. 1.3 days pMRI-cMRI, 2.76 days LKW-pMRI, 2.04 days LKW-cMRI). Although pMRI WMHv correlated with cMRI (r=0.843, p<0.001), pMRI underestimated volumes by 12.3% on average (Bland-Altman 95% LOA -85.9-+61.4%). Log-transformed pMRI WMHv was a strong predictor of discharge mRS (OR 2.34, 95% CI 1.28-4.28, p=0.006) and discharge disposition (OR 2.12, 95% CI 1.10-4.11, p=0.026). Surprisingly, hyperlipidemia had lower adjusted odds of worse outcomes (e.g. mRS OR 0.358, 95% CI 0.179-0.720, p=0.003). In 76 patients with 3-month follow-up, WMHv did not predict 3-month mRS. Hypertension (OR 7.53, 95% CI 1.28-44.4, p=0.026) and diabetes (OR 7.00, 95% CI 1.078-45.4, p=0.041) were independently associated with 3-month outcomes. NIHSS was a strong predictor of all outcomes (e.g. discharge mRS OR 1.15, 95% CI 1.10-1.21, p<0.001).
Conclusions:
pMRI produced WMHv that closely tracked cMRI despite underestimation. pMRI-derived WMHv aids early outcome prediction, while 3-month outcomes are driven by stroke severity and vascular comorbidities.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.