Controllo di switches dinamici nel cervello mediante dispositivi di neuromodulazione minimamente invasivi

A deeper understanding of how neuronal processes and networks are affected by pathologies is essential to achieve insights into brain conditions, and suggest potential routes for treatment. In vivo investigations of neuronal networks encounter formidable challenges, due to a limited observability and a limited possibility of testing manipulation strategies. Neuronal cultures coupled with High-Density Multi-Electrode Arrays (HD-MEAs) provide a well-controlled platform for measuring and modulating network activity through electrical stimulation or pharmacological interventions. Yet, fully leveraging this potential requires robust methods to infer causal links and detect plastic changes from neuronal recordings. In this thesis, we develop a robust methodology to infer Effective Connectivity (EC) in neuronal networks, validate its predictive power for targeted network modulation, and apply it to detect changes induced by repeated stimulation or pharmacological interventions. Following extensive testing on in silico data, we apply three connectivity inference techniques—Transfer Entropy, Cross Correlation, and Cross Covariance—to spontaneous activity data obtained in vitro from dissociated hippocampal neuron cultures derived from rat embryos and plated on high-density microelectrode arrays (HD-MEAs, MaxWell Biosystems). To validate our connectivity estimates, we design and implement an in vitro experimental protocol that tests the causal effects of node stimulation on the rest of the network, providing a "ground truth" for the presence of connections. We also establish the accuracy of EC in identifying optimal stimulation sites for targeted perturbation experiments using direct electrical stimulation. Finally, we employ our inference framework to detect network changes induced by external manipulation, comparing pre- and post-stimulation connectivity in experiments aimed at exploring the effects of Theta Burst Stimulation (TBS) on cultured hippocampal neurons. Through this threefold strategy, the thesis lays the foundation for more precise and causally grounded interventions in neuronal networks, with applications ranging from basic neuroscience to the development of therapeutic protocols.