Neural Noise

Oscillatory brain state-dependent stimulation with TMS combined with EEG

2 min read

  • TMS combined with EEG enables state-dependent neuromodulation by closing the loop between oscillatory brain state and stimulation timing (Nature Protocols, 2026).1
  • Relevant for non-invasive neuromodulation and future EEG-based BCI methods.1
  • Published as a step-by-step protocol by Zrenner, Belardinelli, and Ziemann (University of Toronto / University of Tübingen) in Nature Protocols (DOI: 10.1038/s41596-025-01309-7); the protocol requires ~10 hours to complete and requires standard MRI neuronavigation equipment plus a TMS-compatible EEG amplifier with online output access.1
  • The protocol covers brain MRI and image segmentation for anatomical modeling, accurate EEG source reconstruction, pre-experiment EEG recording to validate oscillation origin and phase targeting accuracy, and the main closed-loop EEG-TMS experiment.1
  • Core principle: cortical excitability fluctuates with ongoing oscillatory phase; TMS delivered at the negative trough of the sensorimotor μ-rhythm (8–13 Hz) yields larger and more consistent motor-evoked potentials (MEPs) than stimulation at the positive peak.2
  • Effective phase-locked stimulation requires end-to-end system latency below ~50–100 ms; real-time algorithms must continuously estimate instantaneous phase via causal band-pass filtering and adaptive prediction, not post-hoc analysis.2
  • Beyond the μ-rhythm, beta (14–30 Hz) and theta bands are also targeted; phase-specific theta-burst TMS over the DLPFC has been shown to accelerate working memory performance, and beta-phase targeting shows an inverted excitability profile relative to mu.3
  • State-dependent TMS induces LTP- or LTD-like plasticity depending on stimulation timing relative to oscillatory phase, opening pathways for personalized therapeutic protocols in depression, stroke, epilepsy, and schizophrenia.2
  • Clinical translation is limited by TMS artifact contamination of EEG, requiring amplifier blanking, sample-and-hold circuits, and ICA-based cleaning; standardization of preprocessing pipelines remains a critical barrier.2
  • Adaptive and machine learning–driven closed-loop systems (reinforcement learning, state-space models) are emerging to dynamically update stimulation timing and intensity based on evolving brain state, potentially enabling individualized precision neuromodulation.2

Footnotes

  1. https://pubmed.ncbi.nlm.nih.gov/41760982/ ↩ ↩2 ↩3 ↩4

  2. https://www.brainview.com/img/pub/pub_47.pdf ↩ ↩2 ↩3 ↩4 ↩5

  3. https://pmc.ncbi.nlm.nih.gov/articles/PMC11528152/ ↩

Graph View