Search

Deep learning technology is uniquely suited to analyse neurophysiological signals such as the electroencephalogram (EEG) and local field potentials (LFP) and promises to outperform traditional machine-learning based classification and feature extraction algorithms. Furthermore, novel cognitive computing platforms such as IBM's recently introduced neuromorphic TrueNorth chip allow for deploying deep learning techniques in an ultra-low power environment with a minimum device footprint. Merging deep learning and TrueNorth technologies for real-time analysis of brain-activity data at the point of sensing will create the next generation of wearables at the intersection of neurobionics and artificial intelligence.

High-fidelity intracranial electrode arrays for recording and stimulating brain activity have facilitated major advances in the treatment of neurological conditions over the past decade. Traditional arrays require direct implantation into the brain via open craniotomy, which can lead to inflammatory tissue responses, necessitating development of minimally invasive approaches that avoid brain trauma. Here we demonstrate the feasibility of chronically recording brain activity from within a vein using a passive stent-electrode recording array (stentrode). We achieved implantation into a superficial cortical vein overlying the motor cortex via catheter angiography and demonstrate neural recordings in freely moving sheep for up to 190 d. Spectral content and bandwidth of vascular electrocorticography were comparable to those of recordings from epidural surface arrays. Venous internal lumen patency was maintained for the duration of implantation. Stentrodes may have wide ranging applications as a neural interface for treatment of a range of neurological conditions.

Directional tuning is the tendency for cortical neurons to exhibit a peak firing rate when a limb is moved in a preferred direction. This phenomenon has been used to underpin decoding strategies in many brain-machine interface (BMI) systems. Although it is well established that individual motor neurons can be decoded using directional tuning, this is not as well understood at the scale of cortical local field potentials (LFPs). This study investigates the directional tuning properties of broadband electrocorticography (ECoG) recorded during a center-out task from two human participants. Selected bipolar ECoG channels demonstrated directional tuning in signal power from 85 – 250 Hz for both subjects. Directional tuning was observed across sensorimotor cortex, as well as frontal areas of cortex. The presence of directional tuning in broadband ECoG suggests the potential use of tuning curves as the basis of a LFP based BMI system.

This work describes a generalized method for classifying motor-related neural signals for a brain-computer interface (BCI), based on a stochastic machine learning method. The method differs from the various feature extraction and selection techniques employed in many other BCI systems. The classifier does not use extensive a-priori information, resulting in reduced reliance on highly specific domain knowledge. Instead of pre-defining features, the time-domain signal is input to a population of multi-layer perceptrons (MLPs) in order to perform a stochastic search for the best structure. The results showed that the average performance of the new algorithm outperformed other published methods using the Berlin BCI IV (2008) competition dataset and was comparable to the best results in the Berlin BCI II (2002–3) competition dataset. The new method was also applied to electroencephalography (EEG) data recorded from five subjects undertaking a hand squeeze task and demonstrated high levels of accuracy with a mean classification accuracy of 78.9% after five-fold cross-validation. Our new approach has been shown to give accurate results across different motor tasks and signal types as well as between subjects.

The expansion of frontiers in neural engineering is dependent on the ability to track, detect and predict dynamics in neural tissue. Recent innovations to elucidate information from electrical recordings of brain dynamics, such as epileptic seizure prediction, have involved switching to an active probing paradigm using electrically evoked recordings rather than traditional passive measurements. This paper positions the advantage of probing in terms of information extraction, by using a coupled oscillator Kuramoto model to represent brain dynamics. While active probing performs better at observing underlying system synchrony in Kuramoto networks, especially in non-Gaussian measurement environments, the benefits diminish with increasing relative size of electrode spatial resolution compared to synchrony area. This suggests probing will be useful for improved characterization of synchrony for suitably dense electrode recordings.