LAUSR

Question Closed-loop methods that trigger the TMS pulse based on the EEG signal are increasingly available, however, current systems rely on the data of a small number of channels and the signal at sensor level cannot be localized to a specific brain area. The challenge is to analyze a sufficient number of EEG channels in real-time to enable spatially localized estimation of individual brain network activity. Methods EEG data is acquired using a high-density TMS compatible recording system and streamed online to a real-time digital processing system based on Simulink. A sliding window of data is used to estimate the sources that give rise to the EEG signal using a spatial filter computation executed on the processor as the signal is acquired. The forward model is computed based on the MRI data using a FEM model to solve the forward problem. The leadfields are calculated employing the neuronavigated electrode positions and then employed to compute the real-time LCMV beamforming. Instantaneous phase is estimated at multiple source-level location simultaneously using parallel sliding windows of band-pass filtered data preceding the current time, each extended into the future in real-time using autoregressive model, with instantaneous phase estimated using a Hilbert transform. Results A latency of Conclusions A novel technique enabling TMS triggered by the instantaneous phase-state of the individual network is presented. This is a significant extension of the current state-of-the-art and could enable the development of more effective personalized closed-loop neuromodulatory stimulation protocols.