INCREASING SLOW-OSCILLATION-SPINDLE COUPLING WITH TEMPORAL INTERFERENCE STIMULATION OF THE THALAMUS IN A COMPUTATIONAL MODEL OF DEEP SLEEP

Coupling between cortical slow oscillations (SOs) and thalamic spindles supports memory consolidation¹. We use subject-specific neural mass models (NMMs), constructed using an MRI analysis pipeline², to investigate the effects of temporal interference stimulation (TIS)³ on SO-spindle coupling. Thalamic nuclei are represented by a previously-published thalamic NMM⁴. We mesh the cortical surface and instantiate a previously-published cortical NMM⁴ at each vertex. Local connectivity is distance-dependent; long-range connectivity is informed by diffusion tensor imaging. TIS is modeled as a driving stimulus proportional to the applied E-field's modulation envelope. The model is implemented in BraiNN, a high-performance whole-brain modelling framework⁵. EEG is simulated by multiplying pyramidal cell output with lead-field matrices calculated in Sim4Life⁶ (Zurich MedTech AG) . SOs and spindles are detected and quantified using YASA⁷.

Without stimulation, the model produced SOs and spindles. Open-loop thalamic TIS produced spindles, but did not generate SOs or increase SO-spindle coupling. However, closed-loop thalamic TIS elicited spindles, increased the amplitude of the SO, and increased SO-spindle coupling. Simultaneous cortical and thalamic stimulation also generated SOs and spindles. These results suggest that TIS holds the potential to improve memory consolidation during sleep. Our findings highlight the role of reciprocal thalamocortical interactions, and demonstrate the utility of closed-loop control and mutli-site stimulation.

[1] King et al. (2017). https://doi.org/10.1016/j.neubiorev.2017.04.026
[2] Karimi et al. (2025). https://doi.org/10.1088/1741-2552/adb88f
[3] Cassara et al. (2025). https://doi.org/10.1002/bem.22536
[4] Costa et al. (2016). https://doi.org/10.1371/journal.pcbi.1005022
[5] Fasse et al. (2025). Paper TH3.R2.120, IEEE NER 2025
[6] Neufeld et al. (2013). https://doi.org/10.1098/rsfs.2012.0058
[7] Vallat et al. (2021). https://doi.org/10.7554/eLife.70092