EEG-guided robotic control of transcranial magnetic stimulation

EEG-derived brain state can jointly control both TMS timing and robotic coil location, with a CNN discriminating resting versus left/right hand movement to select the stimulated hemisphere.

• Why it matters: Existing brain-state-dependent TMS systems adjust only stimulation timing, limiting flexible targeting of distributed cortical networks in closed-loop paradigms.
• What they did: A pre-trained convolutional neural network inferred brain state from 2-s EEG segments, requiring 150 ms for preprocessing/decoding, and drove robotic coil placement that took 7.5 s to reach predefined motor targets.
• The result: In a proof-of-concept study, classifier output determined the hemisphere for primary-motor-cortex hand area stimulation, enabling closed-loop location control based on volitional activity.