Evaluating Unimodal and Multimodal Tracking Strategies for the Reconstruction of Language-related White Matter Tracts

Abstract Purpose Accurate reconstruction of language-related white matter pathways is essential for preoperative planning in brain surgery. While functional (f) MRI activation is often used to guide diffusion (d)MRI tractography, advanced automated protocols rely instead on subcortical anatomical priors. In this study, we evaluate the robustness of anatomically informed protocols without fMRI and compare them with fMRI-guided approaches. Methods Twenty healthy adults (aged 18–32) underwent fMRI during a language task and dMRI on a 3T MRI scanner. Six language-associated fiber bundles were reconstructed with XTRACT using unimodal and multimodal protocols. Reconstruction similarity was assessed using cross-correlations within-subjects, within-cohort, and across-cohorts. Normalized streamline counts served as a proxy for connectivity, and t -tests were used to quantify differences between tracking protocols. Results Protocols using anatomically informed subcortical seed and target masks, as well as their combination with functional masks, yielded higher agreement and greater normalized streamline counts than the fMRI-only protocol. The combined approach showed an additional advantage for reconstructing parieto-temporal white matter tracts. Conclusions These comparisons underscore how protocol choice shapes the reconstruction of language pathways and highlight the need to evaluate these tractography strategies in clinical cohorts.