Personalised therapeutic brain stimulation with closed-loop EEG-synchronized TMS
Our ability to treat pathological brain conditions depends on our ability to interact with brain networks. EEG and transcranial magnetic stimulation (TMS) are non-invasive tools to “read and write” in the language of the brain. We often know where to stimulate, but not when to stimulate, to modulate a dysfunctional circuit. Recent evidence has established “time” as a promising additional target dimension to interact with specific brain networks, expanding the scope of personalized non-invasive brain intervention. In this presentation, the hypothesis will be put forward, that the neuromodulatory effects of TMS depend on the relative timing of the stimulus with respect to fluctuating cortical excitability states. Four open neurophysiological questions will be discussed: (1) The "when" question: What is the optimal phase of a brain oscillation during which a TMS pulse has the strongest neuromodulatory effect? (2) The "where" question: What is the source of the predictive EEG oscillation and how can this best be extracted from the EEG? (3) The "how" question: What are the prerequisites for reliable real-time single-trial estimation of the target brain state? (4) The "what" question: At which intensity and which frequency should how many pulses of TMS be applied to achieve the desired neuromodulatory effect? Recent results will be presented, that show that the phase of the 9-13 Hz sensorimotor rhythm corresponding to the state of highest excitability is in fact not the negative peak, as previously assumed, but the early rising phase, ca. 10 ms later. We will also discuss the data indicating that the most predictive oscillation with regard to pre-central (motor cortex) excitability originates not from the are of cortex targeted by TMS but from the post-central (sensory) cortex. We will then address the fundamental limitations of estimating single-trial oscillatory phase in the presence of noise. The implications of these results, derived from the motor system, for personalized TMS targeting brain networks and locations outside of the motor system will be discussed. Specifically, we will address the question whether the hypothesized neurophysiological mechanisms generalize and may be applied in different protocols designed to modulate dysfunctional brain networks. Finally, we will discuss ongoing and proposed clinical trials of personalized TMS and current challenges of implementing closed-loop therapeutic TMS clinically.