Fig. 1: Overview of the analysis pipeline and the MNMI framework.
a Overview of the analysis pipeline. We use resting-state functional MRI (rs-fMRI) data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and diffusion tensor imaging (DTI) data from the Human Connectome Project (HCP). After preprocessing, functional connectivity (FC) matrices are calculated for the ADNI subjects and structural connectivity (SC) matrices are computed for the HCP subjects. Using FC and SC, MNMI estimates both intra-regional and inter-regional connection strengths for each individual ADNI subject and regional excitation-inhibition (E-I) balance is derived based on the estimated connection strengths. We then perform statistical analysis to identify disrupted connectivity and impaired E-I balance in MCI and AD. Lastly, we examine the association between regional E-I balance and cognitive performance represented by the Mini-Mental State Examination (MMSE) or the Clinical Dementia Rating (CDR) score. b Overview of the MNMI framework. The neural activity (x) is described by a neural mass network model containing multiple brain regions (R1, R2, etc.). Each region consists of one excitatory (E) and one inhibitory (I) neural population coupled with reciprocal connections. The excitatory neural population excites itself (WEE) and the inhibitory neural population (WEI), while receiving feedback inhibition from the latter (WIE). Inter-regional connection strength (W12, W21, W13, etc.) is constrained by structural connectivity from diffusion MRI. The neural activity (x) is converted to the corresponding blood-oxygen-level-dependent (BOLD) signal (y) via a hemodynamic response function and simulated FC is computed. The model parameters are optimized to minimize the difference between simulated FC and empirical FC calculated based on rs-fMRI.
