Fecha de publicación: -- Fuente: PubMed "swarm" Magn Reson Imaging. 2026 Nov 9:110843. doi: 10.1016/j.mri.2026.110843. Online ahead of print.ABSTRACTQuantitative mapping of the myelin water fraction (MWF) is a key objective in neuroimaging, yet the ambiguity of MWF estimates arising from ill-posed inversion problems is rarely assessed at the voxel level. We propose a global-search inversion framework based on particle swarm optimization (PSO) for single and joint inversion of multi-echo MRI data (T2 and T2*) without additional regularization of the individual inverse problems. Repeated PSO runs per voxel yield ensembles of solutions that enable an empirical appraisal of near-optimal solution variability under the chosen model assumptions and misfit criterion. Simulation results show accurate recovery of MWF under the investigated conditions and demonstrate how joint inversion can constrain the solution space. In vivo experiments reveal differences between T2- and T2*-based MWF estimates, with variations across subjects. Joint inversion yields compromise solutions, while preserving sensitivity to inter-contrast discrepancies. The proposed framework complements conventional MWF imaging by providing voxelwise information on the variability of near-optimal solutions, offering a basis for evaluating inversion stability and solution ambiguity.PMID:43003194 | DOI:10.1016/j.mri.2026.110843