PRIN BOBCAT

BOBCAT – Bayesian One-step Brain Connectivity Analysis Tool is a PRIN research project developing new mathematical and computational methods to study functional brain networks from magneto- and electroencephalography (M/EEG) data.

M/EEG provides a unique millisecond view of brain activity, making it possible to investigate how distributed brain regions interact over time. Current approaches to source-space connectivity usually rely on a multi-step workflow: brain activity is first reconstructed from sensor data, then reduced to a smaller set of signals, and finally used to estimate connectivity. These steps require several subjective methodological choices and may introduce artefacts or discard relevant information.

BOBCAT proposes a new one-step Bayesian framework that directly relates the cross-spectrum of the measured M/EEG data to the cross-spectrum of brain sources. The project will develop a sparse model of brain connectivity and an efficient Sequential Monte Carlo algorithm for its estimation. The approach aims to provide more accurate and interpretable connectivity networks, reduce spurious and missed connections, and quantify uncertainty in the estimated networks within a virtually parameter-free framework.

The method will be validated using synthetic data and applied in a pilot study on the identification of epileptogenic networks, with potential relevance for a broad range of neuroscience and clinical applications.