Gideon F. Inbar

An improved P300-based brain-computer interface

A brain-computer interface (BCI) is a system for direct communication between brain and computer. The BCI developed in this work is based on a BCI described by Farwell and Donchin in 1988, which allows a subject to communicate one of 36 symbols presented on a 6 /spl times/ 6 matrix. The system exploits the P300 component of event-related brain potentials (ERP) as a medium for communication. The processing methods distinguish this work from Donchins work. In this work, independent component analysis (ICA) was used to separate the P300 source from the background noise. A matched filter was used together with averaging and threshold techniques for detecting the existence of P300s. The processing method was evaluated offline on data recorded from six healthy subjects. The method achieved a communication rate of 5.45 symbols/min with an accuracy of 92.1% compared to 4.8 symbols/min with an accuracy of 90% in Donchins work. The online interface was tested with the same six subjects. The average communication rate achieved was 4.5 symbols/min with an accuracy of 79.5% as apposed to the 4.8 symbols/min with an accuracy of 56% in Donchins work. The presented BCI achieves excellent performance compared to other existing BCIs, and allows a reasonable communication rate, while maintaining a low error rate.

Effects of muscle model parameter dispersion and multi-loop segmental interaction on the neuromuscular system performance

The effects of parameter dispersion among motor units on the neuromuscular system performance as well as interaction between muscle segments and spinal cord mechanisms are investigated. Elementary components of the system are modeled to simulate with simple models their input-output characteristics. A leaky SS-IPFM encoder with a time-dependent threshold simulates the motor-neuron encoding characteristics. An amplitude and time dependent non-linear model represent the motor unit mechanical output to neuronal input relationship. The dispersion of parameters in the components of the whole muscle control model is investigated in the open loop mode. It is shown that the dispersion of parameters in the multi-efferent channels converging on a common tendon provides a spatial filtration generating a smoother muscle force in addition to extending the linear dynamic range compared to a similar system having identical motor units. Muscle segmental interaction is investigated in this distributed model by closing the loop through a coupling matrix, representing afferent-motorneuron interaction on the spinal cord level. A diagonal matrix represents no segmental interaction and a uniform matrix represents a uniform interaction between segments through the muscle spindles and Golgi tendon feedback elements. The close loop simulation studied shows that (a). The type of segmental interaction has little effect on the overall system performance, i.e., range of linerity and stability, which is the result of having a muscle system with a large number of motor units. (b) There are only minor differences in results between the uniform and normal parameter distributions tested. (c) A loop gain of 4÷8 in the distributed model can provide linearity through the full physiological force range. (d) Type of segmental interaction has significant effects on the individual segment. A uniform matrix provides a more stable segment due to the spatial filtration resulting from the segmental interaction, while the diagonal noninteracting matrix shows instabilities on the local segmental level despite global stability. The more realistic exponentially decaying spatial interaction matrix yields both global neuromuscular and local segmental stability with the same linear dynamic range generated with the uniform or diagonal matrices.

Parameter and Signal Adaptation in the Stretch Reflex Loop

Publisher Summary Using the adaptive system in engineering as a background, a speculative model reference adaptive scheme for the muscle control system is proposed in this chapter. The model is then simulated, using physiological data, to investigate the relative effectiveness and feasibility of parameters and signal adaption. The simulations are followed by experimental results supporting the adaptation concepts in addition to the manifestation of unconscious learning—equivalent to the building of the model reference in the present scheme. The basic results can be summarized as—the parameters of the small signal, linear muscle model are non-linearly and strongly interrelated; the range of variations of the reflex loop parameters is physiologically bound; signal levels in the central nervous system (CNS) have higher bounds than those of the reflex loop; in view of the first two points, it is quite obvious that parameter adaptation schemes developed for engineering purposes are inadequate, and that parameter adaptation is inferior to signal adaptation; and signal adaptation is feasible in cases unaffected by neural delay. Parameter adaptation helps modify the response to external disturbance input, as long as the system is well identified.

Classification of finger activation for use in a robotic prosthesis arm

Hand amputees would highly benefit from a robotic prosthesis, which would allow the movement of a number of fingers. In this paper we propose using the electromyographic signals recorded by two pairs of electrodes placed over the arm for operating such prosthesis. Multiple features from these signals are extracted whence the most relevant features are selected by a genetic algorithm as inputs for a simple classifier. This method results in a probability of error of less than 2%.

Autoregressive Modeling of Surface EMG and Its Spectrum with Application to Fatigue

The following is an investigation of the ability of the autoregressive (AR) model to describe the spectrum of the processes underlying the recorded surface EMG. Surface EMG (SEMG) spectrum is influenced by two major factors; one attributed to the motor units (MU) firing rate and the second, the higher frequency one, to the morphology of the action potentials (AP) traveling along the muscle fiber. In the present paper, SEMG measurements were carried out on the biceps brachii muscle with fixed surface electrodes arrangement and isotonic conditions. Sufficient averaging of 0.5 s segments enabled the identification of the low-frequency peak related to the firing rates of the MUs.

On Surface EMG Spectral Characterization and Its Application to Diagnostic Classification

Surface EMG was recorded from the biceps with fixed muscle length at S0 percent maximal voluntary contraction. The signal bandpass was 10-230 Hz where most of the surface EMG energy is located. The signal sampled at 500 Hz was found to have a changing spectrum. Stationary segments of 500 ms were subject to linear prediction mathematics to model the system.

Muscle spindles in muscle control

Afferent discharges from single muscle spindles in the frog were recorded from the dorsal root ganglion in the reflexively active loop and in the same loop opened by cutting the dorsal roots. The closed loop response (i.e. pulse frequency) to muscle stretch was smooth and without “dead zones”, as compared to the response to the same stretch when the loop was open. The variations in gain between the open and closed loop modes of operation are shown to be small, both for muscle spindle afferent outflow and for muscle tension.

Modeling and estimation of single evoked brain potential components

Presents a novel approach to solving the single-trial evoked-potential estimation problem. Recognizing that different components of an evoked potential complex may originate from different functional brain sites and can be distinguished according to their respective latencies and amplitudes, the authors propose an estimation approach based on identification of evoked potential components on a single-trial basis. The estimation process is performed in 2 stages: first, an average evoked potential is calculated and decomposed into a set of components, with each component serving as a subtemplate for the next stage; then, the single measurement is parametrically modeled by a superposition of an emulated ongoing electroencephalographic activity and a linear combination of latency and amplitude-corrected component templates. Once optimized, the model provides the 2 assumed signal contributions, namely the ongoing brain activity and the single evoked brain response. The estimators performance is analyzed analytically and via simulation, verifying its capability to extract single components at low signal-to-noise ratios typical of evoked potential data. Finally, 2 applications are presented, demonstrating the improved analysis capabilities gained by using the proposed approach. The first application deals with movement related brain potentials, where a change of the single evoked response due to external loading is detected. The second application involves cognitive event-related brain potentials, where a dynamic change of 2 overlapping components throughout the experimental session is detected and tracked.

Feature selection for the classification of movements from single movement-related potentials

Classification of movement-related potentials recorded from the scalp to their corresponding limb is a crucial task in brain-computer interfaces based on such potentials. Many features can be extracted from raw electroencephalographic signals to be used for classification, but the utilization of irrelevant or superfluous features is detrimental to the performance of classification algorithms. It is, therefore, necessary to select a small number of relevant features for the classification task. This paper demonstrates the use of two feature selection methods to choose a small number (10-20) of relevant features from a bank containing upward of 1000 features. One method is based on information theory and the other on the use of genetic algorithms. We show that the former is poorly suited for the aforementioned classification task and discuss the probable reasons for this. However, using a genetic algorithm on data recorded from five subjects we demonstrate that it is possible to differentiate between the movements of two limbs with a classification accuracy of 87% using as little as 10 features without subject training. With the addition of a simple coding scheme, this method can be applied to multiple limb classification and a 63% classification accuracy rate can be reached when attempting to distinguish between three limbs.

The development of a model reference adaptive controller to control the knee joint of paraplegics

Functional neuromuscular stimulation is a technique for restoring motor function by directly activating paralyzed muscles. The design and development of closed-loop controllers that, when used in a simulation system, achieve regulated, repeatable control of the lower limb joints is discussed. Models describing the dynamic behavior of the unloaded lower limb joints under electrical stimulation are described. These models consist of a nonlinear part followed by linear dynamics described by deterministic autoregressive moving average (ARMA) models. These models are used in the design of an adaptive controller to control the movement of the leg joints. Design requirements including a severe constraint on the control rate so as not to excite spastic reflexes, are formulated, and a model reference adaptive controller design which was modified and implemented is discussed. >