Abstract
BACKGROUND: Restrictive diffusion on magnetic resonance imaging is recognized as an early marker of ischemic brain damage, even though diffusion-weighted imaging lesion reversal (DWI-R) is well known. This study aimed to compare methodologies for detecting DWI-R, including voxel-based analysis, which captures subtle lesion dynamics, and to test their correlation with clinical outcomes. METHODS: We retrospectively analyzed magnetic resonance imaging data from 216 consecutive patients with acute ischemic stroke obtained before and after endovascular therapy. DWI-R was defined either as an increase in DWI-Alberta Stroke Program Early Computed Tomography Score, a decrease of total DWI signal volume or as partial reversal of the initial DWI lesion, irrespective of the final DWI load. Associations between 3-month poststroke modified Rankin scale score and DWI-R was assessed according to the different definitions of DWI-R using logistic binary regression. RESULTS: In patients undergoing endovascular therapy, 25% had increased DWI-Alberta Stroke Program Early Computed Tomography Score and 32% showed reduced DWI volume. Both measures were strongly associated with favorable outcomes (modified Rankin Scale score ≤2) with odds ratios of 4.90 and 5.60, respectively (95% CIs: 1.66-14.46 and 2.09-14.98). Voxel-based analysis revealed DWI-R of ≥20% of the initial lesion in 64.5% of cases. Even with an overall increase of lesion volume due to progression elsewhere, ≥20% reversal of initial lesion was associated with a significantly improved outcome compared with <20% reversal, odds ratio 2.22 (95% CIs: 1.05-4.70). CONCLUSION: DWI-R was common in patients treated with endovascular therapy and linked to favorable outcomes. Subtle lesion dynamics detected only by the voxel-based analysis also conferred significant clinical benefits, supporting DWI-R as a continuum rather than a binary measure as "present" or "absent."