Computational Modelling of B₀ Field Strength Effects on Stroke MRI Image Quality: 0.5T vs 1.5T
Abstract
Ischemic stroke outcomes depend on rapid, accurate neuroimaging, and magnetic resonance imaging (MRI) performance is governed by static magnetic field strength. Low-field 0.5 Tesla systems form a substantial share of installed MRI infrastructure in Nigeria and similar resource-limited settings, yet their comparative performance for stroke imaging has not previously been examined using a physics-based computational framework. This study modeled the effects of field strength on MRI image quality in ischemic stroke imaging, comparing 0.5 Tesla and 1.5 Tesla. A digital brain phantom (256 by 256 pixels) was constructed with six tissue compartments, including cerebrospinal fluid and a simulated ischemic stroke lesion in the left middle cerebral artery territory. Magnetic field distributions were computed using a susceptibility-perturbation finite element approximation in Python, followed by spin-echo signal generation, k-space synthesis, and image reconstruction, evaluated using signal-to-noise ratio, contrast-to-noise ratio, normalized mutual information, and image sharpness across thirty noise realizations per field strength. At 1.5 Tesla, signal-to-noise ratio increased by 72.41% and contrast-to-noise ratio by 73.07% relative to 0.5 Tesla, while image sharpness decreased by 32.38% relative to 0.5 Tesla; both metrics changed monotonically and consistently between the two field strengths examined. The simulated stroke lesion satisfied the Rose detectability criterion at both field strengths, with a substantially larger safety margin at 1.5 Tesla (+169.8%) than at 0.5 Tesla (+55.9%). These findings indicate that 1.5 Tesla is the superior field strength for stroke MRI across most evaluated criteria, and the two-point interpolation relationships derived between 0.5T and 1.5T provide a preliminary computational reference for protocol comparison between these two field strengths in resource-limited radiology practice.