Decoding the interplay between brain structural and functional connectivity in Alzheimer’s disease.

The relationship between brain structural (SC) and functional (FC) connectivity changes in Alzheimer’s disease (AD) is crucial for understanding the progression and impact of the disease at different stages. In this study a multivariate statistical technique called Partial Least Squares (PLS) was used to capture the variations in SC and FC between control (CN) and pathological (PAT) subjects. The focus is on the ongoing AD Neuroimaging Initiative (ADNI)-3 cohort, relying on data from diffusion (dMRI) for SC and resting-state functional magnetic resonance imaging (rs-fMRI) for FC. Once standard data preprocessing and PLS were applied, the model weights were visualized based on the functional atlas used. This visualization allowed for a clearer representation of the relationships that were identified. Statistical tests on the original SC and FC data revealed significant differences between the two populations in SC, but not in FC, demonstrating a non-clear correspondence one-to-one between them. However, the PLS model showed unique patterns of SC and FC connections that optimally covary in 13 significant components. Interestingly, anticorrelation among the top weights was found between SC and FC in several components, suggesting an opposite behaviour between these two connectivities in the same links. These findings suggest that PLS can provide a more in-depth understanding of the relationship between SC and FC in AD and highlight the importance of their coupling in determining the pathophysiological pathways of the disease.