Closed-loop Neuroscience of brain rhythms: optimizing real-time quantification of narrow-band signals to expedite feedback delivery

Abstract

AbstractClosed-loop Neuroscience is based on the experimental approach where the ongoing brain activity is recorded, processed, and passed back to the brain as sensory feedback or direct stimulation of neural circuits. The artificial closed loops constructed with this approach expand the traditional stimulus-response experimentation. As such, closed-loop Neuroscience provides insights on the function of loops existing in the brain and the ways the flow of neural information could be modified to treat neurological conditions.Neural oscillations, or brain rhythms, are a class of neural activities that have been extensively studied and also utilized in brain rhythm-contingent (BRC) paradigms that incorporate closed loops. In these implementations, instantaneous power and phase of neural oscillations form the signal that is fed back to the brain.Here we addressed the problem of feedback delay in BRC paradigms. In many BRC systems, it is critical to keep the delay short. Long delays could render the intended modification of neural activity impossible because the stimulus is delivered after the targeted neural pattern has already completed. Yet, the processing time needed to extract oscillatory components from the broad-band neural signals can significantly exceed the period of oscillations, which puts a demand for algorithms that could minimize the delay.We used EEG data collected in human subjects to systematically investigate the performance of a range of signal processing methods in the context of minimizing delay in BRC systems. We proposed a family of techniques based on the least-squares filter design – a transparent and simple approach, as it required a single parameter to adjust the accuracy versus latency trade-off. Our algorithm performed on par or better than the state-of the art techniques currently used for the estimation of rhythm envelope and phase in closed-loop EEG paradigms.