Table of contents

Brain–computer interface (BCI) technology can provide nonmuscular communication and control to people who are severely paralyzed. BCIs can use noninvasive or invasive techniques for recording the brain signals that convey the user's commands. Although noninvasive BCIs are used for simple applications, it has frequently been assumed that only invasive BCIs, which use electrodes implanted in the brain, will be able to provide multidimensional sequential control of a robotic arm or a neuroprosthesis. The present study shows that a noninvasive BCI using scalp-recorded electroencephalographic (EEG) activity and an adaptive algorithm can provide people, including people with spinal cord injuries, with two-dimensional cursor movement and target selection. Multiple targets were presented around the periphery of a computer screen, with one designated as the correct target. The user's task was to use EEG to move a cursor from the center of the screen to the correct target and then to use an additional EEG feature to select the target. If the cursor reached an incorrect target, the user was instructed not to select it. Thus, this task emulated the key features of mouse operation. The results indicate that people with severe motor disabilities could use brain signals for sequential multidimensional movement and selection.

Hindlimb movement in the cat induced by electrical stimulation with an amplitude-modulated waveform of the dorsal surface of the L5-S1 spinal cord or the L5-S1 dorsal/ventral roots was investigated before and after acute spinal cord transection at the T13-L1 level. Stimulation of the spinal cord or dorsal/ventral root at the same spinal segment induced similar movements including coordinated multi-joint flexion or extension. The induced movements changed from flexion to extension when the stimulation was moved from rostral (L5) to caudal (S1) spinal segments. Stimulation of a dorsal or ventral root on one side induced only ipsilateral hindlimb movement. However, stimulation on the dorsal surface of the spinal cord along the midline or across the spinal cord induced bilateral movements. The extension induced by stimulation of L7 dorsal root produced the largest ground reaction force that was strong enough to support body weight. Dorsal root stimulation induced a larger ground reaction force than ventral root stimulation and produced a more graded recruitment curve. Stepping at different speeds could be generated by combined stimulation of the rostral (L5) and the caudal (L6/L7) spinal segments with an appropriate timing between the different stimulation channels. Acute transection of the spinal cord did not change the responses indicating that the induced movements did not require the involvement of the supraspinal locomotor centers. The methods and the stimulation strategy developed in this study might be utilized to restore locomotor function after spinal cord injury.

Using a multi-channel platinum surface electrode array, recordings from cat primary visual cortex were obtained in response to visual stimuli, and electrical stimuli delivered using the elements of the array itself. Neural responses to electrical stimuli were consistent, regardless of stimulus polarity or leading phase (biphasic), although thresholds were lower for monophasic than biphasic pulses. Both visual and electrical stimuli reliably evoked responses with characteristic components, which interacted with each other in a nonlinear summation showing first facilitation then suppression during the window of interaction. The chronaxie for eliciting threshold cortical responses was about 100 µs, and the charge density with a pulse width of 50–100 µs was around 55 µC cm⁻². These data form the basis of understanding the types of cortical responses to stimuli delivered by devices suitable for chronic implantation.

The use of multivariate measurements to characterize brain activity (electrical, magnetic, optical) is widespread. The most common approaches to reduce the complexity of such observations include principal and independent component analyses (PCA and ICA), which are not well suited for discrimination tasks. We addressed two questions: first, how do the neurophysiological responses to elongated phonemes relate to tone and phoneme responses in normal children, and, second, how discriminable are these responses. We employed fully optimized linear discrimination analysis to maximally separate the multi-electrode responses to tones and phonemes, and classified the response to elongated phonemes. We find that discrimination between tones and phonemes is dependent upon responses from associative regions of the brain apparently distinct from the primary sensory cortices typically emphasized by PCA or ICA, and that the neuronal correlates corresponding to elongated phonemes are highly variable in normal children (about half respond with neural correlates of tones and half as phonemes). Our approach is made feasible by the increase in computational power of ordinary personal computers and has significant advantages for a wide range of neuronal imaging modalities.

Urinary retention is the inability to empty the bladder completely, and may result from bladder hypocontractility, increases in outlet resistance or both. Chronic urinary retention can lead to several urological complications and is often refractory to pharmacologic, behavioral and surgical treatments. We sought to determine whether electrical stimulation of sensory fibers in the pudendal nerve could engage an augmenting reflex and thereby improve bladder emptying in an animal model of urinary retention. We measured the efficiency of bladder emptying with and without concomitant electrical stimulation of pudendal nerve afferents in urethane-anesthetized rats. Voiding efficiency (VE = voided volume/initial volume) was reduced from 72 ± 7% to 29 ± 7% following unilateral transection of the sensory branch of the pudendal nerve (UST) and from 70 ± 5% to 18 ± 4% following bilateral transection (BST). Unilateral electrical stimulation of the proximal transected sensory pudendal nerve during distention-evoked voiding contractions significantly improved VE. Low-intensity stimulation at frequencies of 1–50 Hz increased VE to 40–51% following UST and to 39–49% following BST, while high-intensity stimulation was ineffective at increasing VE. The increase in VE was mediated by increases in the duration of distention-evoked voiding bladder contractions, rather than increases in contraction amplitude. These results are consistent with an essential role for pudendal sensory feedback in efficient bladder emptying, and raise the possibility that electrical activation of pudendal nerve afferents may provide a new approach to restore efficient bladder emptying in persons with urinary retention.

People can learn to control EEG features consisting of sensorimotor rhythm amplitudes and can use this control to move a cursor in one or two dimensions to a target on a screen. Cursor movement depends on the estimate of the amplitudes of sensorimotor rhythms. Autoregressive models are often used to provide these estimates. The order of the autoregressive model has varied widely among studies. Through analyses of both simulated and actual EEG data, the present study examines the effects of model order on sensorimotor rhythm measurements and BCI performance. The results show that resolution of lower frequency signals requires higher model orders and that this requirement reflects the temporal span of the model coefficients. This is true for both simulated EEG data and actual EEG data during brain–computer interface (BCI) operation. Increasing model order, and decimating the signal were similarly effective in increasing spectral resolution. Furthermore, for BCI control of two-dimensional cursor movement, higher model orders produced better performance in each dimension and greater independence between horizontal and vertical movements. In sum, these results show that autoregressive model order selection is an important determinant of BCI performance and should be based on criteria that reflect system performance.

We calculated the electric fields induced in the brain during transcranial current stimulation (TCS) using a finite-element concentric spheres human head model. A range of disc electrode configurations were simulated: (1) distant-bipolar; (2) adjacent-bipolar; (3) tripolar; and three ring designs, (4) belt, (5) concentric ring, and (6) double concentric ring. We compared the focality of each configuration targeting cortical structures oriented normal to the surface ('surface-radial' and 'cross-section radial'), cortical structures oriented along the brain surface ('surface-tangential' and 'cross-section tangential') and non-oriented cortical surface structures ('surface-magnitude' and 'cross-section magnitude'). For surface-radial fields, we further considered the 'polarity' of modulation (e.g. superficial cortical neuron soma hyper/depolarizing). The distant-bipolar configuration, which is comparable with commonly used TCS protocols, resulted in diffuse (un-focal) modulation with bi-directional radial modulation under each electrode and tangential modulation between electrodes. Increasing the proximity of the two electrodes (adjacent-bipolar electrode configuration) increased focality, at the cost of more surface current. At similar electrode distances, the tripolar-electrodes configuration produced comparable peak focality, but reduced radial bi-directionality. The concentric-ring configuration resulted in the highest spatial focality and uni-directional radial modulation, at the expense of increased total surface current. Changing ring dimensions, or use of two concentric rings, allow titration of this balance. The concentric-ring design may thus provide an optimized configuration for targeted modulation of superficial cortical neurons.

We have improved the stability and computational efficiency of a physiologically realistic, virtual muscle (VM 3.*) model (Cheng et al 2000 J. Neurosci. Methods101 117–30) by a simpler structure of lumped fiber types and a novel recruitment algorithm. In the new version (VM 4.0), the mathematical equations are reformulated into state-space representation and structured into a CMEX S-function in SIMULINK. A continuous recruitment scheme approximates the discrete recruitment of slow and fast motor units under physiological conditions. This makes it possible to predict force output during smooth recruitment and derecruitment without having to simulate explicitly a large number of independently recruited units. We removed the intermediate state variable, effective length (L_eff), which had been introduced to model the delayed length dependency of the activation–frequency relationship, but which had little effect and could introduce instability under physiological conditions of use. Both of these changes greatly reduce the number of state variables with little loss of accuracy compared to the original VM. The performance of VM 4.0 was validated by comparison with VM 3.1.5 for both single-muscle force production and a multi-joint task. The improved VM 4.0 model is more suitable for the analysis of neural control of movements and for design of prosthetic systems to restore lost or impaired motor functions. VM 4.0 is available via the internet and includes options to use the original VM model, which remains useful for detailed simulations of single motor unit behavior.

Diffuse reflectance spectroscopy as a method for improving intracerebral guidance during functional neurosurgery has been investigated. An optical probe was developed for measurements during stereotactic and functional neurosurgery in man. The aim of the study was to investigate the spectral differences between white and grey matter and between white matter and functional targets. Diffuse reflectance spectroscopy measurements in ten patients were recorded at incremental steps towards and in three different functional targets (STN, GPi and Zi). The recorded spectra along the trajectory were sorted into white or grey matter, based on preoperative MRI images or the recorded spectral shape and intensity. The difference between tissue types was calculated as a quotient. Significant intensity differences between white and grey matter were found to be at least 14% (p < 0.05) and 20% (p < 0.0001) for MRI and spectral-sorted data respectively. The reflectance difference between white matter and the functional targets of GPi was higher than for STN and Zi. The results indicate that diffuse reflectance spectroscopy has a potential to be developed to a suitable complement to other intracerebral guidance methods.

Improving our ability to localize bioelectric sources within a peripheral nerve would help us to monitor the control signals flowing to and from any limb or organ. This technology would provide a useful neuroscience tool, and could perhaps be incorporated into a neuroprosthesis interface. We propose to use measurements from a multi-contact nerve cuff to solve an inverse problem of bioelectric source localization within the peripheral nerve. Before the inverse problem can be addressed, the forward problem is solved using finite element modeling. A fine mesh improves the accuracy of the forward problem solution, but increases the number of variables to be solved for in the inverse problem. To alleviate this problem, variables corresponding to mesh elements that are not distinguishable by the measurement setup are grouped together, thus reducing the dimension of the inverse problem without impacting on the forward problem accuracy. A quantitative criterion for element distinguishability is derived using the columns of the leadfield matrix and information about the uncertainty in the measurements. Our results indicate that the number of variables in the inverse problem can be reduced by more than half using the proposed method, without having a detrimental impact on the quality of the localization.

Polyacrylamide and poly(ethylene glycol) diacrylate hydrogels were synthesized and characterized for use as drug release and substrates for neuron cell culture. Protein release kinetics was determined by incorporating bovine serum albumin (BSA) into hydrogels during polymerization. To determine if hydrogel incorporation and release affect bioactivity, alkaline phosphatase was incorporated into hydrogels and a released enzyme activity determined using the fluorescence-based ELF-97 assay. Hydrogels were then used to deliver a brain-derived neurotrophic factor (BDNF) from hydrogels polymerized over planar microelectrode arrays (MEAs). Primary hippocampal neurons were cultured on both control and neurotrophin-containing hydrogel-coated MEAs. The effect of released BDNF on neurite length and process arborization was investigated using automated image analysis. An increased spontaneous activity as a response to the released BDNF was recorded from the neurons cultured on the top of hydrogel layers. These results demonstrate that proteins of biological interest can be incorporated into hydrogels to modulate development and function of cultured neural networks. These results also set the stage for development of hydrogel-coated neural prosthetic devices for local delivery of various biologically active molecules.

We describe a brain–computer interface for controlling a humanoid robot directly using brain signals obtained non-invasively from the scalp through electroencephalography (EEG). EEG has previously been used for tasks such as controlling a cursor and spelling a word, but it has been regarded as an unlikely candidate for more complex forms of control owing to its low signal-to-noise ratio. Here we show that by leveraging advances in robotics, an interface based on EEG can be used to command a partially autonomous humanoid robot to perform complex tasks such as walking to specific locations and picking up desired objects. Visual feedback from the robot's cameras allows the user to select arbitrary objects in the environment for pick-up and transport to chosen locations. Results from a study involving nine users indicate that a command for the robot can be selected from four possible choices in 5 s with 95% accuracy. Our results demonstrate that an EEG-based brain–computer interface can be used for sophisticated robotic interaction with the environment, involving not only navigation as in previous applications but also manipulation and transport of objects.

Trauma sustained to the central nervous system is a debilitating problem for thousands of people worldwide. Neuronal regeneration within the central nervous system is hindered by several factors, making a multi-faceted approach necessary. Two factors contributing to injury are the irregular geometry of injured sites and the absence of tissue to hold potential nerve guides and drug therapies. Biocompatible hydrogels, injectable at room temperature, that rapidly solidify at physiological temperatures (37 °C) are beneficial materials that could hold nerve guidance channels in place and be loaded with therapeutic agents to aid wound healing. Our studies have shown that thermoreversible methylcellulose can be combined with agarose to create hydrogel blends that accommodate these properties. Three separate novel hydrogel blends were created by mixing methylcellulose with one of the three different agaroses. Gelation time tests show that the blends solidify at a faster rate than base methylcellulose at 37 °C. Rheological data showed that the elastic modulus of the hydrogel blends rapidly increases at 37 °C. Culturing experiments reveal that the morphology of dissociated dorsal root ganglion neurons was not altered when the hydrogels were placed onto the cells. The different blends were further assessed using dissolution tests, pore size evaluations using scanning electron microscopy and measuring the force required for injection. This research demonstrates that blends of agarose and methylcellulose solidify much more quickly than plain methylcellulose, while solidifying at physiological temperatures where agarose cannot. These hydrogel blends, which solidify at physiological temperatures naturally, do not require ultraviolet light or synthetic chemical cross linkers to facilitate solidification. Thus, these hydrogel blends have potential use in delivering therapeutics and holding scaffolding in place within the nervous system.

The intrinsic viscoelastic properties of muscle are central to many theories of motor control. Much of the debate over these theories hinges on varying interpretations of these muscle properties. In the present study, we describe methods whereby a comprehensive musculoskeletal model can be used to make inferences about motor control strategies that would account for behavioral data. Muscle activity and kinematic data from a monkey were recorded while the animal performed a single degree-of-freedom pointing task in the presence of pseudo-random torque perturbations. The monkey's movements were simulated by a musculoskeletal model with accurate representations of musculotendon morphometry and contractile properties. The model was used to quantify the impedance of the limb while moving rapidly, the differential action of synergistic muscles, the relative contribution of reflexes to task performance and the completeness of recorded EMG signals. Current methods to address these issues in the absence of musculoskeletal models were compared with the methods used in the present study. We conclude that musculoskeletal models and kinetic analysis can improve the interpretation of kinematic and electrophysiological data, in some cases by illuminating shortcomings of the experimental methods or underlying assumptions that may otherwise escape notice.

Since sleep is known to influence respiratory activity we studied whether the sleep state would affect the complexity value of the respiratory network output. Specifically, we tested the hypothesis that the complexity values of the diaphragm EMG (EMGdia) activity would be lower during REM compared to NREM. Furthermore, since REM is primarily generated by a homogeneous population of neurons in the medulla, the possibility that REM-related respiratory output would be less complex than that of the awake state was also considered. Additionally, in order to examine the influence of neuron vulnerabilities within the rostral ventral medulla (RVM) on the complexity of the respiratory network output, we inhibited respiratory neurons in the RVM by microdialysis of GABA_A receptor agonist muscimol. Diaphragm EMG, nuchal EMG, EEG, EOG as well as other physiological signals (tracheal pressure, blood pressure and respiratory volume) were recorded from five unanesthetized chronically instrumented intact piglets (3–10 days old). Complexity of the diaphragm EMG (EMGdia) signal during wakefulness, NREM and REM was evaluated using the approximate entropy method (ApEn). ApEn values of the EMGdia during NREM and REM sleep were found significantly (p < 0.05 and p < 0.001, respectively) lower than those of awake EMGdia after muscimol inhibition. In the absence of muscimol, only the differences between REM and wakefulness ApEn values were found to be significantly different.