Subretinal implants are the subject of clinical investigation for their ability to evoke useful visual sensations in blind individuals via electrical stimulation of the diseased retina. We investigated the spatial characteristic of the retinal polarization obtained by electric field stimulation through a subretinally located monopolar electrode array and bipolar electrode array. By combining electric potential simulation through a boundary element method with a segmented cell model, we computed the membrane voltage at the axon terminal of the bipolar cells as a function of the axon length (50–110 $mu {rm m}$) and the electrode diameter. We found that short OFF bipolar cells are predominantly addressed by small bipolar electrodes (diameter between 60 and 100 $ mu{rm m}$) and by using a short duration ($< 150 ~mu {rm s}$) of the stimulating voltage pulse. Long ON cells are best addressed by large monopolar electrodes (diameter $> 100 ~mu {rm m}$) and a long pulse duration ($> 150~ mu {rm s}$). However, the low selectivity of the electric field stimulation with regard to the cell length does not enable the individual depolarization of long OFF cells and short ON cells. When the stimulation must take place at multiple retinal sites simultaneously, the bipolar electrode arrays allow for higher spatial modulation of the polarization of the axon terminal than the monopolar arrays.

In this paper, a novel approach to examine the cortical functional connectivity using multichannel electroencephalographic (EEG) signals is proposed. First we utilized independent component analysis (ICA) to transform multichannel EEG recordings into independent processes and then applied source reconstruction algorithm [i.e., standardize low resolution brain electromagnetic (sLORETA)] to identify the cortical regions of interest (ROIs). Second, we performed a graph theory analysis of the bipartite network composite of ROIs and independent processes to assess the connectivity between ROIs. We applied this proposed algorithm and compared the functional connectivity network properties under resting state condition using 29 student-athletes prior to and shortly after sport-related mild traumatic brain injury (MTBI). The major findings of interest are the following. There was 1) alterations in vertex degree at frontal and occipital regions in subjects suffering from MTBI, $({ p} < 0.05)$; 2) a significant decrease in the long-distance connectivity and significant increase in the short-distance connectivity as a result of MTBI, $({ p} < 0.05)$; 3) a departure from small-world network configuration in MTBI subjects. These major findings are discussed in relation to current debates regarding the brain functional connectivity within and between local and distal regions both in normal controls in pathological subjects.

A relevant issue in a brain–computer interface (BCI) is the capability to efficiently convert user intentions into correct actions, and how to properly measure this efficiency. Usually, the evaluation of a BCI system is approached through the quantification of the classifier performance, which is often measured by means of the information transfer rate (ITR). A shortcoming of this approach is that the control interface design is neglected, and hence a poor description of the overall performance is obtained for real systems. To overcome this limitation, we propose a novel metric based on the computation of BCI Utility. The new metric can accurately predict the overall performance of a BCI system, as it takes into account both the classifier and the control interface characteristics. It is therefore suitable for design purposes, where we have to select the best options among different components and different parameters setup. In the paper, we compute Utility in two scenarios, a P300 speller and a P300 speller with an error correction system (ECS), for different values of accuracy of the classifier and recall of the ECS. Monte Carlo simulations confirm that Utility predicts the performance of a BCI better than ITR.

Sitting acquired pressure ulcers are places of tissue breakdown that mainly occur under the ischial tuberosities (ITs). Successive durations of pressure relief help the buttock tissue recover from sustained deformation and blood-flow stagnation. A computer-aided simulator chair was developed with two adjustable tuberal support elements (TSE) integrated in a force-sensing seating plane (FSP). This study investigated the redistribution of external buttock load in relation to the pattern (i.e., dynamics) of subtuberal blood supply in sitting with a dynamic tuberal support of 1/60 Hz (80 mm/min). Fifteen healthy male subjects were seated with their ITs on the TSE. The experiment involved periodic TSE adjustment in which buttock interface pressure was measured with the FSP and an external pressure mapping device (PMD). Light-guide tissue spectrophotometry was used for simultaneous noninvasive measurement of oxygenation and perfusion in the skin ($<$ 2 mm) and subcutaneous ($<$ 8 mm) tissue under the ITs. TSE adjustment seemed effective to regulate centre of buttock pressure and the forces under the ITs. Differences in measurement with the FSP and PMD have been found due to Hammocking at the seat interface and inaccurate peak pressure readings. Subtuberal blood supply was inversely related to the contact load under the ITs. A rapid inflow of blood in the initial stage of tuberal unloading, followed by a gradual outflow in the rest of the movement cycle indicates that the average blood supply increases when the adjustment frequency increases. Future studies must address the influence of a dynamic tuberal support on the ischial buttock load and pattern of blood supply in impaired individuals.

Gait rehabilitation robots are of increasing importance in neurorehabilitation. Conventional devices are often criticized because they are limited to reproducing predefined movement patterns. Research on patient-cooperative control strategies aims at improving robotic behavior. Robots should support patients only as much as needed and stimulate them to produce maximal voluntary efforts. This paper presents a patient-cooperative strategy that allows patients to influence the timing of their leg movements along a physiologically meaningful path. In this “path control” strategy, compliant virtual walls keep the patient's legs within a “tunnel” around the desired spatial path. Additional supportive torques enable patients to move along the path with reduced effort. Graphical feedback provides visual training instructions. The path control strategy was evaluated with 10 healthy subjects and 15 subjects with incomplete spinal cord injury. The spatio-temporal characteristics of recorded kinematic data showed that subjects walked with larger temporal variability with the new strategy. Electromyographic data indicated that subjects were training more actively. A majority of iSCI subjects was able to actively control their gait timing. Thus, the strategy allows patients to train walking while being helped rather than controlled by the robot.

This paper describes a novel pattern recognition based myoelectric control system that uses parallel binary classification and class specific thresholds. The system was designed with an intuitive configuration interface, similar to existing conventional myoelectric control systems. The system was assessed quantitatively with a classification error metric and functionally with a clothespin test implemented in a virtual environment. For each case, the proposed system was compared to a state-of-the-art pattern recognition system based on linear discriminant analysis and a conventional myoelectric control scheme with mode switching. These assessments showed that the proposed control system had a higher classification error $({ p}<0.001)$ but yielded a more controllable myoelectric control system $({ p}<0.001)$ as measured through a clothespin usability test implemented in a virtual environment. Furthermore, the system was computationally simple and applicable for real-time embedded implementation. This work provides the basis for a clinically viable pattern recognition based myoelectric control system which is robust, easily configured, and highly usable.

We present a new wearable haptic device that provides a sense of position and motion by inducing rotational skin stretch on the user's skin. In the experiments described in this paper, the device was used to provide proprioceptive feedback from a virtual prosthetic arm controlled with myoelectric sensors on the bicep and tricep muscles in 15 able-bodied participants. Targeting errors in blind movements with the haptic device were compared to cases where no feedback and contralateral proprioception were provided. Average errors were lower with the device than with no feedback but larger than with contralateral proprioceptive feedback. Participants also had lower visual demand with the device than with no feedback while tracking a 30 $^circ$ moving range. The results indicate that the rotational skin stretch may ultimately be effective for proprioceptive feedback in myoelectric prostheses, particularly when vision is otherwise occupied.

We present versatile multifunctional programmable controller with bidirectional data telemetry, implemented using existing commercial microchips and standard Bluetooth protocol, which adds convenience, reliability, and ease-of-use to neuroprosthetic devices. Controller, weighing 190 g, is placed on animal's back and provides bidirectional sustained telemetry rate of 500 kb/s , allowing real-time control of stimulation parameters and viewing of acquired data. In continuously-active state, controller consumes $sim$420 mW and operates without recharge for 8 h . It features independent 16-channel current-controlled stimulation, allowing current steering; customizable stimulus current waveforms; recording of stimulus voltage waveforms and evoked neuronal responses with stimulus artifact blanking circuitry. Flexibility, scalability, cost-efficiency, and a user-friendly computer interface of this device allow use in animal testing for variety of neuroprosthetic applications. Initial testing of the controller has been done in a feline model of brainstem auditory prosthesis. In this model, the electrical stimulation is applied to the array of microelectrodes implanted in the ventral cochlear nucleus, while the evoked neuronal activity was recorded with the electrode implanted in the contralateral inferior colliculus. Stimulus voltage waveforms to monitor the access impedance of the electrodes were acquired at the rate of 312 kilosamples/s. Evoked neuronal activity in the inferior colliculus was recorded after the blanking (transient silencing) of the recording amplifier during the stimulus pulse, allowing the detection of neuronal responses within 100 $mu{rm s}$ after the end of the stimulus pulse applied in the cochlear nucleus.

The aim of this study was to describe in detail a new method, called normalized force control parameter (nFCP), to measure changes in movement dynamics obtained during robot-aided neurorehabilitation, and to evaluate its ability to estimate the clinical scales. The study was conducted in a group of 18 subjects after chronic stroke who underwent robot therapy of the upper limb. We used two different measures of movement dynamics to assess patients' performance during each session of training: the nFCP and force directional error (FDE), both measuring the directional error of the patient-exerted force applied to the end-effector of the robot device. Both metrics exhibited significant changes over the three-week course of treatment. The comparison between nFCP and FDE slopes showed a significant and high correlation (${ r} = 0.79$; ${ p} < 0.001$), indicating that the two parameters are closely correlated. The FDE informed on the direction of the force error, while the nFCP showed a better performance in predicting the clinical scale values. Assessment of the time course of recovery showed that nFCP, FDE and the movement smoothness improved quickly at first and then plateaued, while steady gains in mean velocity of movement took place over a longer time course. These data may be helpful to the therapist in developing more effective robot-based therapy protocols.

We have recently demonstrated in simulations and experiments that a proportional and derivative (PD) feedback controller can regulate the active ankle torque during quiet stance and stabilize the body despite a long sensory-motor time delay. The purpose of the present study was to: 1) model the active and passive ankle torque mechanisms and identify their contributions to the total ankle torque during standing and 2) investigate whether a neural-mechanical control scheme that implements the PD controller as the neural controller can successfully generate the total ankle torque as observed in healthy individuals during quiet stance. Fourteen young subjects were asked to stand still on a force platform to acquire data for model optimization and validation. During two trials of 30 s each, the fluctuation of the body angle, the electromyogram of the right soleus muscle, and the ankle torque were recorded. Using these data, the parameters of: 1) the active and passive torque mechanisms (Model I) and 2) the PD controller within the neural-mechanical control scheme (Model II) were optimized to achieve potential matching between the measured and predicted ankle torque. The performance of the two models was finally validated with a new set of data. Our results indicate that not only the passive, but also the active ankle torque mechanism contributes significantly to the total ankle torque and, hence, to body stabilization during quiet stance. In addition, we conclude that the proposed neural-mechanical control scheme successfully mimics the physiological control strategy during quiet stance and that a PD controller is a legitimate model for the strategy that the central nervous system applies to regulate the active ankle torque in spite of a long sensory-motor time delay.

Current brain-computer interface (BCI) systems suffer from high complex feature selectors in comparison to simple classifiers. Meanwhile, neurophysiological and experimental information are hard to be included in these two separate phases. In this paper, based on the hierarchical observation model, we proposed an empirical Bayesian linear discriminant analysis (BLDA), in which the neurophysiological and experimental priors are considered simultaneously; the feature selection, weighted differently, and classification are performed jointly, thus it provides a novel systematic algorithm framework which can utilize priors related to feature and trial in the classifier design in a BCI. BLDA was comparatively evaluated by two simulations of a two-class and a four-class problem, and then it was applied to two real four-class motor imagery BCI datasets. The results confirmed that BLDA is superior in accuracy and robustness to LDA, regularized LDA, and SVM.

We have recently shown that click evoked auditory brainstem responses (ABRs) can be efficiently processed using a novelty detection paradigm. Here, ABRs as a large-scale reflection of a stimulus locked neuronal group synchronization at the brainstem level are detected as novel instance-novel as compared to the spontaneous activity which does not exhibit a regular stimulus locked synchronization. In this paper we propose for the first time Gabor frame operators as an efficient feature extraction technique for ABR single sweep sequences that is in line with this paradigm. In particular, we use this decomposition technique to derive the Gabor frame phase stability (GFPS) of sweep sequences of click and chirp evoked ABRs. We show that the GFPS of chirp evoked ABRs provides a stable discrimination of the spontaneous activity from stimulations above the hearing threshold with a small number of sweeps, even at low stimulation intensities. It is concluded that the GFPS analysis represents a robust feature extraction method for ABR single sweep sequences. Further studies are necessary to evaluate the value of the presented approach for clinical applications.