Retrospective comparison of routine brain MRI scans in patients at 0.55 T and 1.5/3T

To compare the image quality of routine clinical brain MRI sequences at 0.55 T and 1.5/3T and toexplore changes in radiologists' ratings of 0.55 T images over time.

Sixty patients who underwent brain MRI at 0.55 T between October 2021 and April 2022 and September 2022 and October 2022, with scans at 1.5/3T within two years were included. Two neuroradiologists retrospectively evaluated scans from both systems for image quality (IQ) and anatomical features using a 3-point Likert scale (1 = non-diagnostic to 3 = good), assessing the ability to answer clinical questions at 0.55 T. Mean ± SD scores for each sequence were calculated. The paired Wilcoxon test was used to compare IQ between field strengths, while the unpaired Wilcoxon test compared 0.55 T ratings between cohorts. Inter-reader agreement was assessed using linear-weighted Cohen's Kappa coefficient (p ≤ 0.05).

All sequences at 0.55 T were deemed acceptable for diagnostic use (≥2), enabling clinical queries to be addressed despite significant IQ score differences compared to 1.5 T/3T. Scores for FLAIR, SWI (only mIP), T1w TSE, and T2w TSE improved for Reader 1 in the second cohort, while scores for T1w SPACE worsened for Reader 2. Both readers gave consistently low ratings for DWI/ADC, SWI, and contrast-enhanced T1w MPRAGE sequences. Inter-rater agreement improved from moderate (κ = 0.44) in the first cohort to substantial (κ = 0.75) in the second cohort.

Diagnostic-quality brain images can be obtained at 0.55 T, with improved quality scores over time from one reader. Lower ratings for DWI/ADC, SWI, T1w SPACE, and contrast-enhanced T1w MPRAGE sequences suggest that further optimization may be beneficial.