Thalamocortical coupling and cortical E-I balance generate diverse anesthetic α-spindles for interpretable EEG decoding

The neuronal circuits generating frequency (8-13Hz) oscillatory patterns during anesthesia are poorly understood, making their use difficult in predictive medicine. Here, we combine large-scale in vivo recordings of thalamocortical neuronal ensembles in anesthetized mice with a minimal neural mass model. We report non-rhythmic, weakly synchronized firing of thalamic and cortical neurons correlated with oscillatory patterns, indicating an emergent network rather than cellular dynamics. The associated imbalance between the firing of excitatory and inhibitory cortical neurons, together with an increase in firing fluctuations, favors larger amplitudes oscillations, whereas changes in thalamic neuron output control oscillation frequency. A neural mass model, comprising reciprocally connected excitatory and inhibitory cortical and thalamic neurons well reproduces the oscillation patterns observed in vivo at increasing anesthetic doses. These results provide a direct link between EEG oscillations during anesthesia and their underlying network/cellular dynamics which are well captured by a computational model that is clinically relevant for anesthetized brain state monitoring. HighlightsO_LI transient patterns during anesthesia emerge from thalamocortical and cortical excitatory-inhibitory (E-I) network dynamics. C_LIO_LI pattern frequency and amplitude reflect depth of anesthesia, dynamic thalamocortical equilibrium and E-I imbalance. C_LIO_LIEmergence of spindle-like patterns at multiple frequency defines newly characterized brain-state during anesthesia C_LIO_LIA mass-model reproduces the transient nature and spectral diversity of anesthetic-induced oscillations, linking neural circuit mechanisms to EEG output. C_LI