Effect of time windows in LSTM networks for EEG-based BCIs.

People with impaired motor function could be helped by an effective brain-computer interface (BCI) based on a real-time electroencephalogram (EEG) and artificial intelligence algorithms. However, current methodologies for interpreting patient instructions from an EEG are not accurate enough to be completely safe in a real-world situation , where a poor decision would place their physical integrity at risk, such as when traveling in an electric wheelchair in a city. For various reasons, such as the low signal-to-noise ratio of portable EEGs or the effects of signal contamination (disturbances due to user movement, temporal variation of the features of EEG signals, etc.), a long short-term memory network (LSTM) (a type of recurrent neural network) that is able to learn data flow patterns from EEG signals could improve the classification of the actions taken by the user. In this paper, the effectiveness of using an LSTM with a low-cost wireless EEG device in real time is tested, and the time window that maximizes its classification accuracy is studied. The goal is to be able to implement it in the BCI of a smart wheelchair with a simple coded command protocol, such as opening or closing the eyes, which could be executed by patients with reduced mobility. Results show a higher resolution of the LSTM with an accuracy range between 77.61 and 92.14% compared to traditional classifiers (59.71%), and an optimal time window of around 7 s for the task done by users in this work. In addition, tests in real-life contexts show that a trade-off between accuracy and response times is necessary to ensure detection.

A New PLV-Spatial Filtering to Improve the Classification Performance in BCI Systems.

OBJECTIVE: The performance of an EEG-based brain-computer interface (BCI) system is highly dependent on signal preprocessing. This manuscript presents a filtering method to improve the feature classification algorithms typically used in BCI. METHODS: A graph Laplacian quadratic form using the Phase Locking Value (PLV) is applied to generate a new filtered signal in the preprocessing stage. RESULTS: The accuracy of the classification algorithms improved significantly (up to 27.18% in the BCI Competition IV dataset, and up to 42.56% with records made with an Emotiv EPOC+). In addition, the proposed filtering algorithm has similar or better results when compared with the Filter Bank Common Spatial Pattern (FBCSP), which has disadvantages in a multiclass classification. CONCLUSION: This paper shows how our PLV-based filtering between EEG channels could improve the performance of a BCI.

Adaptive pharmacokinetic and pharmacodynamic modelling to predict propofol effect using BIS-guided anesthesia.

BACKGROUND AND OBJECTIVE: Propofol is widely used for hypnosis induction and maintenance of general anesthesia. Its effect can be assessed using the bispectral index (BIS). Many automatic infusion systems are based in pharmacokinetics (PK) and pharmacodynamics (PD) models to predict the response of the patient to the drug. However, all these models do not take into account intra and inter-patient variability. An adjusted intraoperative drug administration allows faster recovery and provides post-operative side-effect mitigation METHODS: BIS evolution and surgery-recorded propofol infusion data of a group of 60 adult patients (30 males/30 females) with ASA I/II physical status were used to test a real time PK/PD compartmental model. This new algorithm tunes three model parameters (ce50, γ and ke0), minimizing a performance function online. RESULTS: The error in the BIS signal predicted by the real time PK/PD model was smaller than the error measured with fixed parameter equations. This model shows that ce50, γ and ke0 change with time and patients, given a mean (95% confidence interval) of 3.89 (3.52-4.26)mg/l, 4.63 (4.13-5.13) and 0.36 (0.31-0.4)min(-1), respectively. CONCLUSIONS: The real time PK/PD model proposed provides a closer description of the patient real state at each sample time. This allows for greater control of the drug infusion, and thus the quantity of drug administered can be titrated to achieve the desired effect for the desired duration, and reduce unnecessary waste or post-operative effects.

Application of active electrode compensation to perform continuous voltage-clamp recordings with sharp microelectrodes.

OBJECTIVE: Electrophysiological recordings of single neurons in brain tissues are very common in neuroscience. Glass microelectrodes filled with an electrolyte are used to impale the cell membrane in order to record the membrane potential or to inject current. Their high resistance induces a high voltage drop when passing current and it is essential to correct the voltage measurements. In particular, for voltage clamping, the traditional alternatives are two-electrode voltage-clamp technique or discontinuous single electrode voltage-clamp (dSEVC). Nevertheless, it is generally difficult to impale two electrodes in a same neuron and the switching frequency is limited to low frequencies in the case of dSEVC. We present a novel fully computer-implemented alternative to perform continuous voltage-clamp recordings with a single sharp-electrode. APPROACH: To reach such voltage-clamp recordings, we combine an active electrode compensation algorithm (AEC) with a digital controller (AECVC). MAIN RESULTS: We applied two types of control-systems: a linear controller (proportional plus integrative controller) and a model-based controller (optimal control). We compared the performance of the two methods to dSEVC using a dynamic model cell and experiments in brain slices. SIGNIFICANCE: The AECVC method provides an entirely digital method to perform continuous recording and smooth switching between voltage-clamp, current clamp or dynamic-clamp configurations without introducing artifacts.