Exploring Microstructural And Network Mechanisms in Stroke using Diffusion MRI and Correlation with Behavioral Recovery

Stroke result in neurological impairment due to both local damage and widespread impact on connectivity organization. Whether alterations of microstructural and structural connectivity patterns follow the same trajectories described for functional connectivity, both in terms of reorganization and correlation with behavior is unclear. For the first project, we investigated a multivariate assessment of microstructural parameters including outcomes from diffusion tensor imaging (DTI) and Neurite orientation dispersion and density imaging (NODDI) models; first, we assessed whether applying this new model we could identify local and distal changes after stroke and then we assessed the relationship between microstructural parameters and behavioral performance in stroke patients. In the second project, we used a gradient methodology to measure longitudinal large-scale changes of white matter organization in stroke, and the relationship with behavioral performance. We performed a prospective study on first-time stroke patients, at two-week and three-month intervals. Patients underwent extensive behavioral assessments and diffusion weighted imaging to assess microstructure and structural connectivity. A latent factorial analysis was applied to behavioral data. For the first project, we used a novel approach for estimating diffusion latent space, utilizing factorial analysis on DTI and NODDI metrics within the human connectome project (HCP) dataset and a separate in our stroke cohort. We defined three specific region of interests (ROIs) for stroke patients: lesional, proximal perilesional, and distal perilesional spaces. In continue, we conducted a longitudinal analysis to examine the time's effect on diffusion factors in lesion, perilesional, and distal regions. Finally, a relationship between cognitive performance and diffusion factors was explored. For the second project, the structural gradients were assessed for both intra- and inter-hemispheric connections and averaged across network templates. Statistical analyses included cross-sectional analysis of variance and longitudinal assessments by means of linear mixed model. Finally, we explored the relationships between structural gradients, behavioral performance, and changes over time. For first project, three main latent diffusion factors were identified within the HCP cohort and independently verified in the our cohort of healthy, demonstrating consistent patterns across groups and individual levels. In the acute phase, significant differences were observed in the white matter (WM) lesion and proximal perilesional space when compared to healthy control, indicating altered microstructure properties in these areas. Longitudinally, significant changes were noted in the WM of the proximal perilesional space, suggesting increase of disintegration over time from acute to chronic stages. Furthermore, a significant link was established between perilesional diffusion properties and motor deficits. These findings highlighted the potential of diffusion metrics as indicators of motor function post-stroke and highlight the dynamic changes in WM brain microstructure following a stroke. For second project, We described three main gradients for both intra- and inter-hemispheric connections. Control sample results were consistent across two time points. Network-wise analysis unveiled widespread acute stage alterations, impacting several networks in ipsi- and contralesional hemispheres. Longitudinal assessment demonstrated significant time effects. Finally, we reported an association between acute structural connectivity patterns and visuospatial-memory performance only at the acute stage, decoupled at the chronic follow-up. Stroke profoundly influences structural connectivity patterns, extending beyond the lesion site. These findings indicate that structural patterns do not recover over time, in contrast to the functional connectome, but rather diverge further from the normative connectome.