Contralesional Brain-Computer Interface Control of a Powered Exoskeleton for Motor Recovery in Chronic Stroke Survivors.

BACKGROUND AND PURPOSE: There are few effective therapies to achieve functional recovery from motor-related disabilities affecting the upper limb after stroke. This feasibility study tested whether a powered exoskeleton driven by a brain-computer interface (BCI), using neural activity from the unaffected cortical hemisphere, could affect motor recovery in chronic hemiparetic stroke survivors. This novel system was designed and configured for a home-based setting to test the feasibility of BCI-driven neurorehabilitation in outpatient environments. METHODS: Ten chronic hemiparetic stroke survivors with moderate-to-severe upper-limb motor impairment (mean Action Research Arm Test=13.4) used a powered exoskeleton that opened and closed the affected hand using spectral power from electroencephalographic signals from the unaffected hemisphere associated with imagined hand movements of the paretic limb. Patients used the system at home for 12 weeks. Motor function was evaluated before, during, and after the treatment. RESULTS: Across patients, our BCI-driven approach resulted in a statistically significant average increase of 6.2 points in the Action Research Arm Test. This behavioral improvement significantly correlated with improvements in BCI control. Secondary outcomes of grasp strength, Motricity Index, and the Canadian Occupational Performance Measure also significantly improved. CONCLUSIONS: The findings demonstrate the therapeutic potential of a BCI-driven neurorehabilitation approach using the unaffected hemisphere in this uncontrolled sample of chronic stroke survivors. They also demonstrate that BCI-driven neurorehabilitation can be effectively delivered in the home environment, thus increasing the probability of future clinical translation. CLINICAL TRIAL REGISTRATION: URL: http://www.clinicaltrials.gov. Unique identifier: NCT02552368.

Cortical adaptation to a chronic micro-electrocorticographic brain computer interface.

Brain-computer interface (BCI) technology decodes neural signals in real time to control external devices. In this study, chronic epidural micro-electrocorticographic recordings were performed over primary motor (M1) and dorsal premotor (PMd) cortex of three macaque monkeys. The differential gamma-band amplitude (75-105 Hz) from two arbitrarily chosen 300 µm electrodes (one located over each cortical area) was used for closed-loop control of a one-dimensional BCI device. Each monkey rapidly learned over a period of days to successfully control the velocity of a computer cursor. While both cortical areas contributed to success on the BCI task, the control signals from M1 were consistently modulated more strongly than those from PMd. Additionally, we observe that gamma-band power during active BCI control is always above resting brain activity. This suggests that purposeful gamma-band modulation is an active process that is obtained through increased cortical activation.

Electrocorticographic (ECoG) correlates of human arm movements.

Invasive and non-invasive brain-computer interface (BCI) studies have long focused on the motor cortex for kinematic control of artificial devices. Most of these studies have used single-neuron recordings or electroencephalography (EEG). Electrocorticography (ECoG) is a relatively new recording modality in BCI research that has primarily been built on successes in EEG recordings. We built on prior experiments related to single-neuron recording and quantitatively compare the extent to which different brain regions reflect kinematic tuning parameters of hand speed, direction, and velocity in both a reaching and tracing task in humans. Hand and arm movement experiments using ECoG have shown positive results before, but the tasks were not designed to tease out which kinematics are encoded. In non-human primates, the relationships among these kinematics have been more carefully documented, and we sought to begin elucidating that relationship in humans using ECoG. The largest modulation in ECoG activity for direction, speed, and velocity representation was found in the primary motor cortex. We also found consistent cosine tuning across both tasks, to hand direction and velocity in the high gamma band (70-160 Hz). Thus, the results of this study clarify the neural substrates involved in encoding aspects of motor preparation and execution and confirm the important role of the motor cortex in BCI applications.

Motor cortical representation of hand translation and rotation during reaching.

Previous studies have shown that hand translation is well represented by neuronal activity in the proximal arm area of primary motor cortex (M1). However, little is known about cortical representation of hand rotation in M1. In this study, single-unit activity was recorded from monkeys while they performed a "center-out with rotation" task. When reaching for a target, subjects had to match four separate kinematic parameters: three-dimensional location and one-dimensional orientation of the target. Among the 512 neurons modulated by hand movement, 446 were tuned to hand translation, 326 were tuned to hand rotation, and 260 neurons were tuned to both hand translation and hand rotation. Approximately half of the neurons that encoded both translation and rotation did so in a nonlinear manner. This nonlinear interaction can be modeled as a gain-field type of encoding whereby hand rotational velocity modulated the hand translational cosine tuning curves in a multiplicative manner. Furthermore, this study demonstrated that both hand translation and hand rotation can be decoded simultaneously from a population of motor cortical neurons.

Proceedings of the Second International Workshop on Advances in Electrocorticography.

The Second International Workshop on Advances in Electrocorticography (ECoG) was convened in San Diego, CA, USA, on November 11-12, 2010. Between this meeting and the inaugural 2009 event, a much clearer picture has been emerging of cortical ECoG physiology and its relationship to local field potentials and single-cell recordings. Innovations in material engineering are advancing the goal of a stable long-term recording interface. Continued evolution of ECoG-driven brain-computer interface technology is determining innovation in neuroprosthetics. Improvements in instrumentation and statistical methodologies continue to elucidate ECoG correlates of normal human function as well as the ictal state. This proceedings document summarizes the current status of this rapidly evolving field.

Defining Surgical Terminology and Risk for Brain Computer Interface Technologies.

With the emergence of numerous brain computer interfaces (BCI), their form factors, and clinical applications the terminology to describe their clinical deployment and the associated risk has been vague. The terms "minimally invasive" or "non-invasive" have been commonly used, but the risk can vary widely based on the form factor and anatomic location. Thus, taken together, there needs to be a terminology that best accommodates the surgical footprint of a BCI and their attendant risks. This work presents a semantic framework that describes the BCI from a procedural standpoint and its attendant clinical risk profile. We propose extending the common invasive/non-invasive distinction for BCI systems to accommodate three categories in which the BCI anatomically interfaces with the patient and whether or not a surgical procedure is required for deployment: (1) Non-invasive-BCI components do not penetrate the body, (2) Embedded-components are penetrative, but not deeper than the inner table of the skull, and (3) Intracranial -components are located within the inner table of the skull and possibly within the brain volume. Each class has a separate risk profile that should be considered when being applied to a given clinical population. Optimally, balancing this risk profile with clinical need provides the most ethical deployment of these emerging classes of devices. As BCIs gain larger adoption, and terminology becomes standardized, having an improved, more precise language will better serve clinicians, patients, and consumers in discussing these technologies, particularly within the context of surgical procedures.

Sensory Percepts Elicited by Chronic Macro-Sieve Electrode Stimulation of the Rat Sciatic Nerve.

Objective: Intuitive control of conventional prostheses is hampered by their inability to provide the real-time tactile and proprioceptive feedback of natural sensory pathways. The macro-sieve electrode (MSE) is a candidate interface to amputees' truncated peripheral nerves for introducing sensory feedback from external sensors to facilitate prosthetic control. Its unique geometry enables selective control of the complete nerve cross-section by current steering. Unlike previously studied interfaces that target intact nerve, the MSE's implantation requires transection and subsequent regeneration of the target nerve. Therefore, a key determinant of the MSE's suitability for this task is whether it can elicit sensory percepts at low current levels in the face of altered morphology and caliber distribution inherent to axon regeneration. The present in vivo study describes a combined rat sciatic nerve and behavioral model developed to answer this question. Approach: Rats learned a go/no-go detection task using auditory stimuli and then underwent surgery to implant the MSE in the sciatic nerve. After healing, they were trained with monopolar electrical stimuli with one multi-channel and eight single-channel stimulus configurations. Psychometric curves derived by the method of constant stimuli (MCS) were used to calculate 50% detection thresholds and associated psychometric slopes. Thresholds and slopes were calculated at two time points 3 weeks apart. Main Results: For the multi-channel stimulus configuration, the average current required for stimulus detection was 19.37 µA (3.87 nC) per channel. Single-channel thresholds for leads located near the nerve's center were, on average, half those of leads located near the periphery (54.92 µA vs. 110.71 µA, or 10.98 nC vs. 22.14 nC). Longitudinally, 3 of 5 leads' thresholds decreased or remained stable over the 3-week span. The remaining two leads' thresholds increased by 70-74%, possibly due to scarring or device failure. Significance: This work represents an important first step in establishing the MSE's viability as a sensory feedback interface. It further lays the groundwork for future experiments that will extend this model to the study of other devices, stimulus parameters, and task paradigms.

Brain-controlled interfaces: movement restoration with neural prosthetics.

Brain-controlled interfaces are devices that capture brain transmissions involved in a subject's intention to act, with the potential to restore communication and movement to those who are immobilized. Current devices record electrical activity from the scalp, on the surface of the brain, and within the cerebral cortex. These signals are being translated to command signals driving prosthetic limbs and computer displays. Somatosensory feedback is being added to this control as generated behaviors become more complex. New technology to engineer the tissue-electrode interface, electrode design, and extraction algorithms to transform the recorded signal to movement will help translate exciting laboratory demonstrations to patient practice in the near future.

Fibrin matrices with affinity-based delivery systems and neurotrophic factors promote functional nerve regeneration.

Glial-derived neurotrophic factor (GDNF) and nerve growth factor (NGF) have both been shown to enhance peripheral nerve regeneration following injury and target different neuronal populations. The delivery of either growth factor at the site of injury may, therefore, result in quantitative differences in motor nerve regeneration and functional recovery. In this study we evaluated the effect of affinity-based delivery of GDNF or NGF from fibrin-filled nerve guidance conduits (NGCs) on motor nerve regeneration and functional recovery in a 13 mm rat sciatic nerve defect. Seven experimental groups were evaluated consisting of GDNF or NGF and the affinity-based delivery system (DS) within NGCs, control groups excluding the DS and/or growth factor, and nerve isografts. Groups with growth factor in the conduit demonstrated equivalent or superior performance in behavioral tests and relative muscle mass measurements compared to isografts at 12 weeks. Additionally, groups with GDNF demonstrated greater specific twitch and tetanic force production in extensor digitorum longus (EDL) muscle than the isograft control, while groups with NGF produced demonstrated similar force production compared to the isograft control. Assessment of motor axon regeneration by retrograde labeling further revealed that the number of ventral horn neurons regenerating across NGCs containing GDNF and NGF DS was similar to the isograft group and these counts were greater than the groups without growth factor. Overall, the GDNF DS group demonstrated superior functional recovery and equivalent motor nerve regeneration compared to the isograft control, suggesting it has potential as a treatment for motor nerve injury.

Local field potentials for BCI control.

The gold standard in brain-computer interface (BCI) modalities is multi single-unit recordings in the primary motor cortex. It yields the fastest and most elegant control (i.e., most degrees of freedom and bitrate). Unfortunately, single-unit electrodes are prone to encapsulation, which limit their single-unit recording life. However, encapsulation does not significantly affect intracortical local field potentials (LFPs). LFPs and single-unit activity were recorded from the motor cortices of three monkeys (Macaca fascicularis) while they performed a standard 3D center-out reaching task and a 3D circle-drawing task. The high frequency (HF) (60-200 Hz) spectral amplitudes of a subset of the LFPs were found to be directionally tuned much like single units. In fact, stable isolation of single units on the same electrode increased the likelihood that the HF-LFP would be significantly cosine tuned to hand direction. The presence of significantly tuned single units further increased the likelihood of a tuned HF-LFP, suggesting that this band of HF-LFP activity is at least partially generated by local neuronal action potential currents (i.e., single-unit activity). Given that encapsulation makes recording single units over a long period of time difficult, these results suggest that HF-LFPs may be a more stable and efficient method of monitoring neural activity for BCI applications.

Conductive Core-Sheath Nanofibers and Their Potential Application in Neural Tissue Engineering.

We have prepared conductive core-sheath nanofibers via a combination of electrospinning and aqueous polymerization. Specifically, nanofibers electrospun from poly(ε-caprolactone) (PCL) and poly((L)-lactide) (PLA) were employed as templates to generate uniform sheaths of polypyrrole (PPy) via in situ polymerization. These conductive core-sheath nanofibers offer a unique system for studying the synergistic effect of different cues on neurite outgrowth in vitro. We found that explanted dorsal root ganglia (DRG) adhered well to the conductive core-sheath nanofibers and generated neurites across the surface when there was a nerve growth factor in the medium. Furthermore, the neurites could be oriented along one direction and enhanced by 82% in terms of maximum length when uniaxially aligned conductive core-sheath nanofibers are compared with their random counterparts. Electrical stimulation, when applied through the mats of conductive core-sheath nanofibers, was found to further increase the maximum length of neurite for random and aligned samples by 83% and 47%, respectively, relative to the controls without electrical stimulation. Combined together, these results suggest the potential use of the conductive core-sheath nanofibers as scaffolds in applications such as neural tissue engineering.

Evolution of brain-computer interfaces: going beyond classic motor physiology.

The notion that a computer can decode brain signals to infer the intentions of a human and then enact those intentions directly through a machine is becoming a realistic technical possibility. These types of devices are known as brain-computer interfaces (BCIs). The evolution of these neuroprosthetic technologies could have significant implications for patients with motor disabilities by enhancing their ability to interact and communicate with their environment. The cortical physiology most investigated and used for device control has been brain signals from the primary motor cortex. To date, this classic motor physiology has been an effective substrate for demonstrating the potential efficacy of BCI-based control. However, emerging research now stands to further enhance our understanding of the cortical physiology underpinning human intent and provide further signals for more complex brain-derived control. In this review, the authors report the current status of BCIs and detail the emerging research trends that stand to augment clinical applications in the future.

Micro-channel sieve electrode for concurrent bidirectional peripheral nerve interface. Part A: recording.

OBJECTIVE: Advancement in prosthetic limb technology requires corresponding improvements in the capability of the amputee to naturally control the device via original motor pathways while simultaneously receiving haptic feedback via sensory pathways. Recording efferent axonal activity using a peripheral neural interface (PNI) allows a good tradeoff between invasiveness and selectivity while possibly preserving the phenomenology of controlling the original limb. One such PNI, the thin-film transverse intrafascicular multichannel electrode (tfTIME), has been shown to be successful in controlling powered prosthetics. However, the tfTIME is highly susceptible to stimulation artifact; thus, using such a PNI to both record efferent motor signals while concurrently stimulating afferent sensory axons in the same nerve is problematic. The micro-channel sieve electrode could also provide a stable, selective, neural interface with larger signal-to-noise levels that are less susceptible to concurrent stimulation artifact or other external noise effects. APPROACH: This study uses a computational model to compare recording levels of simulated ENGs across neural drive levels as well as basic control signals derived from the ENGs in both tfTIME and micro-channel sieve PNIs. A motor neuron pool model generated axon firing rates at a given neural drive. The time course of the corresponding extracellular currents of the myelinated motor axons were determined using core conductor axon models. Finite element models determined the contribution of the extracellular current from nodes of Ranvier on potentials recorded using each interface. Contributions from each node were combined to create the final ENG. MAIN RESULTS: ENGs recorded using the micro-channel sieves were shown to have much higher amplitudes compared to ENGs recorded using the tfTIMEs. Signal amplitudes also varied less as a function of axonal placement and spike timing, resulting in more consistent signals with amplitudes determined predominantly by neural drive. SIGNIFICANCE: Simulation results suggest that the micro-channel sieve provides higher quality control signals over tfTIME PNIs in decoding ENGs. Coupling these results with concurrent stimulation results of the companion paper (Part B: stimulation) suggests that the micro-channel sieve is an optimal bidirectional PNI.

Micro-channel sieve electrode for concurrent bidirectional peripheral nerve interface. Part B: stimulation.

OBJECTIVE: Successful use of a prosthetic limb by an amputee is facilitated by haptic feedback-both a sense of touch and proprioception. Stimulating afferent fibers within peripheral nerves has been shown to provide somatosensation enabling amputees to modulate the control of prosthetic limbs. Peripheral nerve interfaces (PNIs) have also been used to decode patients' motor intentions. It seems ideal to use PNIs to record efferent fibers for motor control while stimulating afferent fibers to create concurrent sensory feedback. However, while many PNIs claim to be bi-directional, few can both stimulate and record at the same time due to stimulation artifacts which are orders of magnitude larger than the recorded motor signals. This study uses computational modelling to compare the stimulation artifact at threshold levels of stimulation for thin-film transverse intrafascicular multichannel electrodes (tfTIMEs) with micro-channel sieve electrodes. APPROACH: Finite element models of micro-channel sieves and tfTIMESs were used to solve for electric fields generated during peripheral nerve stimulation. Electrophysiological responses were simulated using axon models. Stimulation artifacts were calculated for stimuli eliciting axonal action potentials. Simulations were carried out for multiple micro-channel geometries and electrode configurations. MAIN RESULTS: Stimulation artifacts generated for threshold stimulation currents are lower for micro-channel devices compared to tfTIMEs. Consequently, stimulus artifacts at threshold currents were substantially higher for the tfTIME. Micro-channel width has a moderate impact on recruitment thresholds and stimulus artifacts. Using the micro-channel sieve in bipolar and tripolar stimulation configurations greatly decreases stimulation artifacts particularly for optimized contact placements (CPs). Electroneurogram (ENG) signals from the companion paper were incorporated showing a great improvement in signal-to-artifact ratio for the micro-channel electrode compared to tfTIMEs. SIGNIFICANCE: Stimulating regenerated nerve tissue using micro-channel sieve electrodes can decrease stimulation artifacts and elicit neuronal responses at low stimulation amplitudes. Further analysis provides clues to optimal implementations of micro-channel devices. Finally, stimulation artifacts for simulated tfTIME devices were 2-3 orders of magnitude larger than ENG levels. In contrast, for some micro-channel configurations stimulation artifacts were 3-4 orders of magnitude smaller than ENG levels.

Spectral changes in cortical surface potentials during motor movement.

In the first large study of its kind, we quantified changes in electrocorticographic signals associated with motor movement across 22 subjects with subdural electrode arrays placed for identification of seizure foci. Patients underwent a 5-7 d monitoring period with array placement, before seizure focus resection, and during this time they participated in the study. An interval-based motor-repetition task produced consistent and quantifiable spectral shifts that were mapped on a Talairach-standardized template cortex. Maps were created independently for a high-frequency band (HFB) (76-100 Hz) and a low-frequency band (LFB) (8-32 Hz) for several different movement modalities in each subject. The power in relevant electrodes consistently decreased in the LFB with movement, whereas the power in the HFB consistently increased. In addition, the HFB changes were more focal than the LFB changes. Sites of power changes corresponded to stereotactic locations in sensorimotor cortex and to the results of individual clinical electrical cortical mapping. Sensorimotor representation was found to be somatotopic, localized in stereotactic space to rolandic cortex, and typically followed the classic homunculus with limited extrarolandic representation.

Modelling the impact of altered axonal morphometry on the response of regenerative nervous tissue to electrical stimulation through macro-sieve electrodes.

OBJECTIVE: Regenerated peripheral nervous tissue possesses different morphometric properties compared to undisrupted nerve. It is poorly understood how these morphometric differences alter the response of the regenerated nerve to electrical stimulation. In this work, we use computational modeling to explore the electrophysiological response of regenerated and undisrupted nerve axons to electrical stimulation delivered by macro-sieve electrodes (MSEs). APPROACH: A 3D finite element model of a peripheral nerve segment populated with mammalian myelinated axons and implanted with a macro-sieve electrode has been developed. Fiber diameters and morphometric characteristics representative of undisrupted or regenerated peripheral nervous tissue were assigned to core conductor models to simulate the two tissue types. Simulations were carried out to quantify differences in thresholds and chronaxie between undisrupted and regenerated fiber populations. The model was also used to determine the influence of axonal caliber on recruitment thresholds for the two tissue types. Model accuracy was assessed through comparisons with in vivo recruitment data from chronically implanted MSEs. MAIN RESULTS: Recruitment thresholds of individual regenerated fibers with diameters >2 µm were found to be lower compared to same caliber undisrupted fibers at electrode to fiber distances of less than about 90-140 µm but roughly equal or higher for larger distances. Caliber redistributions observed in regenerated nerve resulted in an overall increase in average recruitment thresholds and chronaxie during whole nerve stimulation. Modeling results also suggest that large diameter undisrupted fibers located close to a longitudinally restricted current source such as the MSE have higher average recruitment thresholds compared to small diameter fibers. In contrast, large diameter regenerated nerve fibers located in close proximity of MSE sites have, on average, lower recruitment thresholds compared to small fibers. Utilizing regenerated fiber morphometry and caliber distributions resulted in accurate predictions of in vivo recruitment data. SIGNIFICANCE: Our work uses computational modeling to show how morphometric differences between regenerated and undisrupted tissue results in recruitment threshold discrepancies, quantifies these differences, and illustrates how large undisrupted nerve fibers close to longitudinally restricted current sources have higher recruitment thresholds compared to adjacently positioned smaller fibers while the opposite is true for large regenerated fibers.

Anterior hip joint force increases with hip extension, decreased gluteal force, or decreased iliopsoas force.

Abnormal or excessive force on the anterior hip joint may cause anterior hip pain, subtle hip instability and a tear of the acetabular labrum. We propose that both the pattern of muscle force and hip joint position can affect the magnitude of anterior joint force and thus possibly lead to excessive force and injury. The purpose of this study was to determine the effect of hip joint position and of weakness of the gluteal and iliopsoas muscles on anterior hip joint force. We used a musculoskeletal model to estimate hip joint forces during simulated prone hip extension and supine hip flexion under four different muscle force conditions and across a range of hip extension and flexion positions. Weakness of specified muscles was simulated by decreasing the modeled maximum force value for the gluteal muscles during hip extension and the iliopsoas muscle during hip flexion. We found that decreased force contribution from the gluteal muscles during hip extension and the iliopsoas muscle during hip flexion resulted in an increase in the anterior hip joint force. The anterior hip joint force was greater when the hip was in extension than when the hip was in flexion. Further studies are warranted to determine if increased utilization of the gluteal muscles during hip extension and of the iliopsoas muscle during hip flexion, and avoidance of hip extension beyond neutral would be beneficial for people with anterior hip pain, subtle hip instability, or an anterior acetabular labral tear.

Human perception of electrical stimulation on the surface of somatosensory cortex.

Recent advancement in electrocorticography (ECoG)-based brain-computer interface technology has sparked a new interest in providing somatosensory feedback using ECoG electrodes, i.e., cortical surface electrodes. We conducted a 28-day study of cortical surface stimulation in an individual with arm paralysis due to brachial plexus injury to examine the sensation produced by electrical stimulation of the somatosensory cortex. A high-density ECoG grid was implanted over the somatosensory and motor cortices. Stimulation through cortical surface electrodes over the somatosensory cortex successfully elicited arm and hand sensations in our participant with chronic paralysis. There were three key findings. First, the intensity of perceived sensation increased monotonically with both pulse amplitude and pulse frequency. Second, changing pulse width changed the type of sensation based on qualitative description provided by the human participant. Third, the participant could distinguish between stimulation applied to two neighboring cortical surface electrodes, 4.5 mm center-to-center distance, for three out of seven electrode pairs tested. Taken together, we found that it was possible to modulate sensation intensity, sensation type, and evoke sensations across a range of locations from the fingers to the upper arm using different stimulation electrodes even in an individual with chronic impairment of somatosensory function. These three features are essential to provide effective somatosensory feedback for neuroprosthetic applications.

Computational model of a primate arm: from hand position to joint angles, joint torques and muscle forces.

Three-dimensional reaching by non-human primates is an important behavioral paradigm for investigating representations existing in motor control areas of the brain. Most studies to date have correlated neural activity to a few of the many arm motion parameters including: global hand position or velocity, joint angles, joint angular velocities, joint torques or muscle activations. So far, no single study has been able to incorporate all these parameters in a meaningful way that would allow separation of these often highly correlated variables. This paper introduces a three-dimensional, seven degree-of-freedom computational musculoskeletal model of the macaque arm that translates the coordinates of eight tracking markers placed on the arm into joint angles, joint torques, musculotendon lengths and finally into an optimized prediction of muscle forces. This paper uses this model to illustrate how the classic center-out reaching task used by many researchers over the last 20 years is not optimal in separating out intrinsic, extrinsic, kinematic and kinetic variables. However, by using the musculoskeletal model to design and test novel behavioral movement tasks, a priori, it is possible to disassociate the myriad of movement parameters in motor neurophysiological reaching studies.

Local field potential spectral tuning in motor cortex during reaching.

In this paper, intracortical local field potentials (LFPs) and single units were recorded from the motor cortices of monkeys (Macaca fascicularis) while they preformed a standard three-dimensional (3-D) center-out reaching task. During the center-out task, the subjects held their hands at the location of a central target and then reached to one of eight peripheral targets forming the corners of a virtual cube. The spectral amplitudes of the recorded LFPs were calculated, with the high-frequency LFP (HF-LFP) defined as the average spectral amplitude change from baseline from 60 to 200 Hz. A 3-D linear regression across the eight center-out targets revealed that approximately 6% of the beta LFPs (18-26 Hz) and 18% of the HF-LFPs were tuned for velocity (p-value < 0.05), while 10% of the beta LFPs and 15% of the HF-LFPs were tuned for position. These results suggest that a multidegree-of-freedom brain-machine interface is possible using high-frequency LFP recordings in motor cortex.