Towards brain state classification with functional ultrasound imaging

This study investigates the potential of functional ultrasound imaging (fUSI) as a promising, non-invasive, and cost-effective tool for identifying brain states and detecting pathological changes. We simulated brain activity using a Wilson-Cowan mass model, converting electrophysiological signals into fUSI data through a hemodynamic response function (HRF). By introducing pathological alterations to the model, we assessed fUSI’s ability to distinguish between healthy (control) and diseased (Alzheimer’s disease, epilepsy) conditions using a neural network classifier. Based on 100 realizations per condition, classification on simulated fUSI signals remained high for Alzheimer’s disease (80% accuracy vs. 95% on electrophysiology), but was lower for epilepsy (65% vs. 95%), likely due to the fast, transient nature of epileptic activity. Nonetheless, our findings show that fUSI can reliably detect distinct brain states in most cases, offering exciting prospects for its use in real-time neurological assessments.