A computational study of visual evoked potentials dynamics in dementia with a mesoscale brain model

The main goal of my work is to replicate the results obtained from an experimental setup. In this experiment, individuals from three different categories, namely healthy subjects (HS), subjective cognitive decline (SCD), and mild cognitive impairment (MCI), were asked to perform a visual attention task. Here different geometrical figures were displayed and the corresponding electrophysiological response to the stimulus, event-related potentials (ERPs), were recorded. The analysis of EEG data, coming from channels in correspondence of the visual cortex, revealed alterations in visual evoked potentials (VEPs) which correspond to the first physiological response to the stimulus and that are nonlinear with the worsening of the cases considered. To be specific, the negative N1 component of the evoked potential, having its minimum after 80-120 ms from the onset of the stimuli, was found to be more pronounced in the CTR case, but less pronounced in SCD compared to in MCI. However, the causal mechanism of this behaviour at network level is still unclear. To investigate further, a phenomenological mesoscale brain model was used to reproduce, from the simulated brain structure, the electrophysiological behaviour observed analytically. To do so, simulations have been performed via The Virtual Brain (TVB) software, which allows to infer in a model-based way the neurophysiological behaviours, reproducing brain dynamics and generating simulated neuroimaging signals such as EEG, functional MRI and MEG. The whole brain structure is depicted by 76 nodes, each representing a brain region and described by a set of stochastic differential equations (SDE). The mathematical framework used in order to simulate the EEG activity is the one developed by Jansen and Rit, describing the activity of three different neural populations: two excitatory and one inhibitory. The main result was to be able in reproducing the experimental VEPs tuning different model parameters to reproduce a structural degeneration. The ones concerning the brain network dynamics were modified, specifically those of neurons activity and the ones concerning the connectome. Moreover, a reduction in the stimulus amplitude was found to be able in reproducing the ordering of the N1 peaks observed empirically. Subsequently, these results were extended across the whole dementia continuum. This was done considering steps in between the two extremes obtaining intermediate scenarios in between those considered. The trend seems to be oscillating around the fixed values of the conditions considered. Findings suggest structural or functional alterations in the path from retina to V1, which affects either lateral geniculate nucleus, or thalamus, or even the retina itself. Thus, the stimulus could already be attenuated upon reaching the visual cortex. This type of attenuation could be influenced by generic aging problems on retina-V1 path, linked to other pathologies as well coexisting with Alzheimer¿s dementia, and its prodromal stages. These outcomes pave the way for the suggesting of causal mechanisms on N1 VEP dynamics, which might be probed with in vivo investigation of retina-V1 path integrity for early diagnosis.