Functional Neural Activity Mapping Using Spiking Neural Networks and EEG Signals: a Proof of Concept Study

The characterization of the intrinsic neural activity underlying neurophysiological recordings, which can be gathered through noninvasive techniques, is a major goal of computational neuroscience research. While different methods have been proposed to solve the inverse problem from a pure electromagnetic standpoint, literature regarding functional neural reconstruction modeling is limited. This study introduces a novel framework to define and quantify brain activity from a functional perspective, combining spiking neural networks with Electroencephalography (EEG) signal analysis. Single neuron dynamics is described via the Izhikevich model, and each channel activity is modeled as the outcome of a distinct population of cortical inhibitory and excitatory neurons. Functional interactions among distinct populations are also modeled. We validate this framework by testing it with a dataset of real recordings from 30 healthy subjects undergoing a cold pressure test. Our findings reveal a global enhancement of neural spiking activity during the elicitation session, especially in the β and γ bands. Results suggest that the proposed model is capable of describing the underlying functional neural activity of brain dynamics and showing significant variability between a resting state session and the cold pressure session. The proposed method paves the way for a functional alternative of brain source localization problems.