Introduction: Cocaine is one of the most commonly used addictive drugs with adverse medical consequences. Its exposure involves rapid neurodynamic changes that are challenging to capture in awake subjects. This study aimed to characterize acute cocaine-induced alterations in neuronal activity and functional connectivity (FC) using high-resolution functional ultrasound (fUS) and to develop a machine-learning-based framework for state classification. Methods: Using fUS with superior spatiotemporal resolution to fMRI, we measured cerebral blood volume (CBV) as a proxy for neuronal activity and assessed FC in awake mice. A Support Vector Machine (SVM) classifier was trained to distinguish between baseline and cocaine-exposed states using FC patterns across the cortex, hippocampus, and thalamus. Results: Acute cocaine significantly increased cortical CBV, indicating elevated neuronal activation. Concurrently, marked reductions in FC were observed between the cortex, hippocampus, and thalamus-key regions governing cognition and behavior. The SVM classifier reliably differentiated baseline and cocaine-exposed states based on these disrupted FC signatures. Discussion: This approach demonstrates the utility of integrated neuroimaging and AI for probing dynamic brain states in substance use disorders. While providing unprecedented temporal resolution, future studies should address chronic exposure models and include both sexes to enhance translational relevance. Conclusion: fUS combined with machine learning reveals acute cocaine-induced dysregulation of cortico-hippocampal-thalamic networks, identifying disrupted connectivity as a neural substrate of addiction.
Zhou et al. (Mon,) studied this question.