Lee E. Miller - US grants

The ability to reach toward, grasp, and manipulate objects is nearly unique among primates. The loss of this ability, through stroke or other neurological disease is devastating. Despite increasing knowledge of the number of brain areas participating in the guidance of limb movement, several hypotheses remain regarding the role even of the "primary" motor cortex, a structure that has been recognized for over 100 years, and sends neurons directly to the spinal cord to control movement. The proposed experiments seek to explore the nature of the signals within the primary motor cortex, and the manner in which they relate to muscle activation and the control of arm and hand movements. When a person decides to grasp an object, the motor command is expressed perceptually as a desired movement of the end-point of the limb to a particular point in space. Much of the problem faced by the brain arises from the need to convert this signal into a set of muscle activation signals. Data suggest that the primary motor cortex sends a desired hand movement signal to the spinal cord, and only there is it converted to muscle signals. Other data indicate that the signals sent to the spinal cord have already been converted to muscle activation signals within the motor cortex, or at some earlier stage of processing. Signals will be recorded from the motor cortex and limb muscles of monkeys during limb movement. Hand movement, posture of the limb, and joint torques will also be determined. Varied hand use, altered limb posture, and added inertial loads will be used to decrease the covariation among these signals, and correlation and regression methods will be used to determine their relation to the neuronal signals. In later experiments, signals from several neurons will be used to predict the time course of putative output signals. Motor cortical neurons projecting to the spinal cord will be distinguished from those projecting elsewhere. A neuron with discharge that is consistently related to the activity of a particular set of muscles across the varied types of behavior would be evidence for motor commands in muscle coordinates. On the other hand, discharge that consistently encodes either end-point movement or joint coordinates across tasks, would support the idea that motor cortical command signals are expressed in one of these other coordinate systems. We will attempt to relate functionally distinct groups of neurons to information about their cortical location or output projections.

DESCRIPTION (provided by applicant): It has been recognized for hundreds of years, that motor deficits following injury to the brain occur on the same side as a cerebellar lesion, but opposite the side of an injury to the cerebrum. The nature of the deficit also differs. Cerebral cortical injury causes muscle weakness or paralysis, while cerebellar lesions cause uncoordinated movements that are typically of the wrong size. Recognition of these specific motor signs caused great advances in the diagnosis and surgical treatment of such injury, but the personal and societal costs of treatment and rehabilitation following stroke or other injury remain tremendous.There are extensive interconnections between the cerebellum and the cerebral cortex. In fact, the primary motor cortex sends many more fibers to the cerebellum than it does to the spinal cord. Both structures have been studied extensively by means of recordings in behaving animals. Even so, conflicting ideas remain about the relation between these two areas, and the way in which they interact to produce and refine motor command signals. We propose to examine these relations by recording simultaneously in both the cerebellar nuclei (CN) and primary motor cortex (M1) in the awake, behaving monkey. In addition to paired recordings, we propose to use electrical micro-stimulation and averaging methods to examine the interconnections, and to micro-inject drugs during recording in order to distinguish convergent inputs. These measurements will be made during normal movement, and during motor learning, as the monkey adapts to novel forces imposed bya robotic device, that affect the trajectory of its movements.Simultaneous recordings provide far more than the ability to do two experiments at once. Recording simultaneously from select pairs of cells under literally identical experimental conditions provides a level of spatial and temporal precision with which to compare response properties that is not possible with single site recordings. More importantly, we will be able to associate these properties with the underlying functional connections between particular pairs of recording sites. This approach should greatly enhance our ability to understand the relation between M 1 and CN during the generation of motor command signals.

DESCRIPTION (provided by applicant): In a scant few years, the Brain Machine Interface (BMI) has gone from science fiction to a scientific curiosity, to a rapidly growing engineering discipline with real potential for clinical importance. The realization that signals recorded from the brain might be used to control inanimate objects has captured the fascination of the popular and scientific communities alike. However, despite tremendous increase in attention and scientific work, two fundamental limitations remain: 1) The great majority of BMIs extract only kinematic (position) information from the brain, ignoring the wealth of force-related information that is also present in the primary motor cortex, and 2) virtually all existing BMIs depend exclusively on natural vision to guide movement, lacking the rapid proprioceptive feedback that is critical for normal movement. We propose to address both of these limitations by building on the progress we have made in the previous grant cycle. We previously demonstrated both joint torque and EMG predictions with accuracy comparable to that of kinematic predictions. We now propose to use this information as the basis both for a torque-based controller, and an adaptive, hybrid torque-position controller. The decoder will use inputs from both primary motor cortex and the dorsal premotor cortex. We hypothesize that this approach will allow the monkey subjects to perform more realistic tasks that require movement in a changing and changing dynamical environment. Two typical examples are the need to grasp and move an object, and the need to control both endpoint force and position, for example, when writing. We have also demonstrated that visually guided BMI performance can be improved with the addition of natural proprioception, and that monkeys can discriminate electrical stimuli of different intensity in proprioceptive areas of the cortex. We now propose to stimulate these areas to provide artificial proprioceptive feedback to the monkey. We will stimulate particular electrodes with patterns intended to mimic the signals that occur when the monkey's limb is perturbed during the movement. We hypothesize that the stimulation will cause the monkey to initiate a short latency correction in a direction determined by the particular characteristics of the stimulation. Ultimately we propose to combine the hybrid, adaptive controller with the proprioceptive prosthesis, and to test the monkey's ability to adapt to the two interfaces. We postulate that that plastic changes in the cortex, combined with algorithmic adaptation will drive improvements in performance with a time course of several days to a week. The proposed experiments will lead directly to clearer understandings of the signals encoded in both the motor and sensory areas of the brain, and the adaptive processes that are critical when a patient recovers from neurological and musculoskeletal disorders like stroke, amputation, or spinal cord injury. Furthermore, we anticipate that the developed technology will directly benefit these same patients as it is moved from the experimental arena to the clinic.

[unreadable] DESCRIPTION (provided by applicant): The goal of this project is to develop a primate model of an upper extremity neuromuscular stimulation system controlled by means of intra-cortical recording electrodes. Individuals with spinal cord injury become paralyzed because they have lost the ability to activate their muscles. These patients' muscles can still be made to contract if they are activated by means of electrical stimuli applied directly to the muscle or nerves. Likewise, the areas of the brain that normally control movement are still active, but their connection to the muscles has been lost as a result of the injury. Researchers at Case Western Reserve University (CWRU) have demonstrated that implanted functional electrical stimulation (FES) neuroprostheses can be used to restore grasp functions to individuals with tetraplegia. Although remarkable, these systems are limited to pre-programmed grasp patterns, and require considerable conscious attention. A more natural control system, with more degrees of freedom could provide greatly improved function. At Northwestern, we have developed methods to predict the activity of arm and hand muscles during grasping movements based on micro-electrode recordings from the brain of a monkey. From a single, chronically implanted array of electrodes, predictions can be made of the activity of shoulder, arm and hand muscles. This type of electrode has yielded maintained recordings for periods in excess of 3 years, and it has recently been approved for experimental use in human patients. We believe that intra-cortical recordings like these provide the potential for simultaneous control of multiple degrees of freedom through natural thought processes. By combining the strengths of the Northwestern and CWRU groups, we propose to develop a brain-computer interface adequate for controlling a neuroprosthesis. The development of a primate model of this neuroprosthetic system would be a major step toward its implementation in human patients. This application includes the following specific aims: 1) We propose to use a 100-electrode array implanted in the primary motor cortex of a mnkey to provide the input to a set of decoders designed to produce real-time predictions of the activity of particular hand muscles. 2) We propose to use the control algorithms developed in aim 1 and an implanted FES prosthesis to restore grasp following temporary muscle paralysis induced by a pharmacological nerve block. 3) We propose to develop these control algorithms without the use of initial EMG measurements, as would be necessary in order to implement the system for a patient. [unreadable] [unreadable]

DESCRIPTION (provided by applicant): When asked, most spinal cords injured (SCI) patients suffering tetraplegia report that regaining the ability to use their hands would be more important than any other lost function. Functional Electrical Stimulation (FES) is a remarkable technology that can be used to cause the muscles of a paralyzed patient to contract. FES has been used to restore grasping to paralyzed patients. The primary limitations of FES for the restoration of dexterous hand movements are the inability to activate muscles with adequate force and specificity, and the inadequacy of the control signals that the paralyzed patient can generate. The Brain Machine Interface (BMI) may provide the necessary control signal. BMIs have reached the forefront of the neural engineering endeavor to treat patients suffering from paralysis. Yet despite remarkable BMI technology advances, virtually all current applications are limited to daily, several-hour sessions in a single, constrained setting. Current BMIs that restore movement do so only through a robot or limb exoskeleton, and none provides explicit control of force. These constraints will ultimately limit patients' ability to adapt fully to the technology, to use it readily at any time of the day or night, and to apply it to a broad range of behaviors. By coupling BMI technology to FES, we believe we can overcome these limitations. We have demonstrated a unique BMI using a peripheral nerve block to paralyze a monkey's hand as a model for SCI: We use information about intended muscle activity extracted from cortical recordings to produce an FES control signals that allows the monkeys to regain voluntary control of their wrist and hand. We propose to use this BMI-controlled FES model to restore round-the-clock hand use to monkey subjects for month-long periods of time. We will develop adaptive, state-dependent decoders designed to broaden the range of motor behaviors for which the FES BMI will be useful. We will improve both the quality of information we can obtain from the brain and the effectiveness with which we can activate muscles by using new types of electrodes for neural recording and peripheral nerve stimulation. Finally, we will develop a long-lasting peripheral nerve block to cause month-long paralysis. We will record telemetrically, to control a fully implanted neuromuscular stimulator that will allow us an unprecedented opportunity to study long-term adaptation to a BMI neuroprosthesis. We will study the behavioral improvement that results from this adaptation both in the monkey's natural home-cage behaviors and in the more constrained lab setting. We will study the interaction between the monkey's adaptation and the adaptive algorithms. This work will provide important basic information about the adaptive capability of the adult, mammalian brain, the extent to which BMI exposure can rescue cortex that undergoes maladaptive changes in response to paralysis, and the extent to which long- term practice improves BMI performance.

? DESCRIPTION (provided by applicant): The cuneate nucleus (CN) of the brainstem receives ascending input from somatosensory afferents that innervate the skin, joints, and muscles as well as descending input from sensorimotor cortex. While the tactile and proprioceptive response properties of primary afferents and of neurons in primary somatosensory cortex (S1) have been extensively characterized, virtually nothing is known of the response properties of CN neurons. Multiple types of cutaneous afferents innervate the glabrous skin of the hand and play different albeit overlapping roles in tactile perception. Tactile afferents exhibit highly patterned, repeatable responses when their receptive fields are touched. Likewise, receptors in muscles and tendons provide signals related to muscle length, length change, and force. Within S1, the properties of neurons have been extensively characterized and are very different from their peripheral counterparts. First, S1 neurons receive convergent input from multiple somatosensory submodalities. Second, neurons in S1 convey high-level, processed information about stimulus features (edge orientation, surface texture, direction of limb movement). Third, S1 responses can be strongly modulated by the behavioral state of the animals. We propose to characterize, for the first time, the response properties of CN neurons in awake primates and assess (1) the degree to which signals from different somatosensory submodalities converge onto single CN neurons; (2) the extent to which the feature selectivity observed in S1 begins to emerge in CN; (3) the degree to which the state-dependence of S1 responses stems from CN. We anticipate that the present study will have important implications for our understanding of somatosensory processing and may inform the development of subcortical interfaces for brain-machine interfaces used to restore somatosensation in patients with spinal cord injury or who have lost a limb.

Spinal cord injury causes both paralysis and loss of sensation from the limbs. The past 15 years have seen remarkable advances in ?Brain Machine Interfaces? (BMIs) that allow paralyzed persons to move anthropomorphic limbs using signals recorded directly from their brains. However, these movements remain slow, clumsy, and effortful, looking remarkably like those of individuals who have lost sensation from their arms due to peripheral neuropathy. Brain-controlled prosthetic limbs are unlikely to achieve high levels of performance in the absence of artificial sensory feedback. Early attempts at restoring somatosensation used intracortical microstimulation (ICMS) to activate somatosensory cortex (s1), requiring animals to learn largely arbitrary patterns of stimulation to represent two or three virtual objects or to navigate in two-dimensional space. While an important beginning, this approach seems unlikely to scale to the broad range of limb movements and interactions with objects that we experience in daily life. To move the field past this hurdle, we propose to replace both touch and proprioception by using multi- electrode ICMS to produce naturalistic patterns of neuronal activity in S1 of monkeys. In Aim 1, we will develop model-optimized mappings between limb state (pressure on the fingertip, or motion of the limb) and the patterns of ICMS required to evoke S1 activation that mimics that of natural inputs. These maps will account for both the dynamics of neural responses and the biophysics of ICMS. We anticipate that this biomimetic approach will evoke intuitive sensations that require little or no training to interpret. We will validate the maps by comparing natural and ICMS-evoked S1 activity using novel hardware that allows for concurrent ICMS and neural recording. In Aim 2, we will test the ability of monkeys to recognize objects using artificial touch. Having learned to identify real objects by touch, animals will explore virtual objects with an avatar that shadows their own hand movements, receiving artificial touch sensations when the avatar contacts objects. We will test their initial performance on the virtual stereognosis task without learning, as well as their improvements in performance over time. Aim 3 will be similar, but will focus on proprioception. We will train monkeys to report the direction of brief force bumps applied to their hand. After training, we will replace the actual bumps with virtual bumps created by patterned ICMS, again asking the monkeys to report their perceived sense of the direction and magnitude of the perturbation. Finally, in Aim 4, we will temporarily paralyze the monkey's arm, thereby removing both touch and proprioception, mimicking the essential characteristics of a paralyzed patient. The avatar will be controlled based on recordings from motor cortex and guided by artificial somatosensation. The monkey will reach to a set of virtual objects, find one with a particular shape, grasp it, and move it to a new location. If we can demonstrate that this model-optimized, biomimetic feedback is informative and easy to learn, it should form the basis for robust, scalable, somatosensory feedback for BMIs.

PROJECT SUMMARY Movement is essential for enabling us to interact with the world. All behaviors, including speech, writing, reaching, grasping, walking and posture require the coordinated activities of many motor areas. Further, sensory signals provide essential feedback to these motor areas, enabling accurate motor control and motor learning, as well as providing information vital for deciding future behaviors. As a result, understanding the sensorimotor control of even seemingly simple movements like reaching out to pick up a glass of water is complex. Damage to these sensorimotor pathways can produce a wide range of debilitating neurological disorders including tremor, Parkinson's disease, ataxia, dystonia, and spasticity - all of which markedly decrease patient quality of life. The Society for the Neural Control of Movement (NCM) is an international community of scientists, clinician-investigators and students all engaged in research whose common goal is to understand how the brain controls movement in health, and to address the deficits that occur in disease. NCM promotes a broad range of research using interdisciplinary approaches (e.g., neurophysiological, anatomical, molecular, computational, and behavioral), different animal models, and studies of normal subjects and patients with neurological disorders. The inaugural NCM Meeting took place in 1991. The success of the society and its annual meeting has led to a continual growth in membership, meeting attendance, and the breadth of scientific content. With support through the NIH, the 2018 NCM meeting will make substantive progress towards furthering three main goals of the society: Aim 1) Stimulate new research approaches and collaborations among NCM meeting attendees by identifying new topics and appropriate scientists as speakers, Aim 2) continue to increase the gender and ethnic diversity within the NCM leadership and in meeting programing, and Aim 3) promote and support the development of the next generation of motor control researchers by providing financial and career support for graduate students and post-doctoral fellows. Overall, the unique format of the annual NCM meeting, with its focus on interdisciplinary approaches, discussion, and scientific interaction in an intimate meeting environment, is of immeasurable value to furthering worldwide understanding of how the brain controls movement in both health and disease.

For decades, neuroscientists have recorded from single brain cells (neurons) to understand how the brain senses, makes decisions, and controls movements. We can now record from hundreds of neurons simultaneously but are still at an early stage in developing tools for determining how networks of neurons work together to perceive the world and to generate the control signals needed to produce coordinated movement. Focusing on movement, this project brings to bear the power of deep learning --- powerful new machine learning algorithms --- on the problem of understanding neural activity. Because deep learning thrives on big data, the investigators can leverage massive-scale brain recordings. These include month-long recordings chronicling the activity of 100 neurons as a monkey goes about its daily business, or recording from thousands of neurons for hours in the mouse, each identified with an exact location in the brain and tied to the mouse's on-going behaviors. These approaches will open new windows on how neurons act together moment-by-moment to produce movement. The investigators will develop simple descriptions of the underlying processes to be shared with the public through venues including online tutorials, a new open course that will be developed at Emory University and Georgia Tech, the Atlanta Science Festival, and Atlanta's Brain Awareness Month. They will also make their data sets publicly available, and host data tutorial and modeling competitions at key scientific meetings, to accelerate progress by engaging the broader scientific community.

In the fifty years since Ed Evarts first recorded single neurons in M1 of behaving monkeys, great effort has been devoted to understanding the relation between these individual signals and movement-related signals collected during highly constrained motor behaviors performed by over-trained monkeys. In parallel, theoreticians posited that the computations performed in the brain depend critically on network-level phenomena: dynamical laws in brain circuits that constrain the activity and dictate how it evolves over time. The goal of this project is to develop a powerful new suite of tools, based on deep learning, to analyze these dynamics at unprecedented temporal and spatial scales. The investigators will leverage recordings with month-long M1 electrophysiology, EMG, and behavioral data during natural behaviors from monkeys, and vast numbers of neurons recorded with two-photon imaging from behaving mice. Novel machine learning techniques using sequential auto-encoders will enable the investigators to learn the dynamics underlying these data. This combination will provide windows into the brain's control of motor behavior that have never before been possible. The novel analytical framework developed here will be extensible from motor behaviors to higher level problems of error processing, decision making, and learning.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Abstract Chronically implanted microelectrode arrays (MEAs) for recording extracellular neural activity are central to scientific studies of neural circuit function in behaving animals. These studies seek to understand how neurons encode information and how neural signals can be decoded to provide insights into brain learning and dysfunction. The majority of microelectrode MEAs, and especially those available commercially, are fabricated from silicon and leverage techniques associated with integrated circuit microfabrication and packaging. For microelectrode MEAs, the major limitations for chronic neural recording are the reactive tissue response that encapsulates electrodes and kills or damages neurons in the vicinity of the electrode and the degradation and failure of materials used in MEA fabrication. An effective means of minimizing the foreign body response is the use of ultramicroelectrode MEAs (UMEAs) with subcellular cross-sectional dimensions. In related work, we have demonstrated that carbon-fiber ultramicroelectrodes substantially evade a foreign body response and have been shown to provide stable chronic neural recordings in small-animal models. However, a scalable manufacturing process for carbon-fiber ultramicroelectrodes has not emerged. The proposed effort is aimed at developing and demonstrating the chronic stability and reliability of ultramicroelectrodes based on amorphous silicon carbide (a- SiC) UMEAs that are fabricated by industry-standard thin-film processes. We aim to develop a fabrication process for a-SiC UMEAs with 32 to 128 ultramicroelectrodes and demonstrate the stability of these UMEAs through accelerated laboratory testing and their functionality by neural recording and histology using chronic implants in rat cortex and in a 3-4 year non-human primate study. To facilitate dissemination of the a-SiC UMEA technology, electrical interconnect hardware and implantation methods will also be developed. We anticipate the proposed a-SiC UMEAs impacting the neuroscience community by providing a highly stable neural interface that allows single-unit and ensemble recording for probing neuronal circuitry on a dimensional scale that is not possible with current multielectrode recording devices. We expect a-SiC UMEAs will provide new insights into the neural networks and changes in neural circuitry that may accompany behavior and adaption.