Extensive literatures have shown approaches for decoding upper limb kinematics or muscle activity using multichannel cortical spike recordings toward brain machine interface (BMI) applications. However, similar topics regarding lower limb remain relatively scarce. We previously reported a system for training monkeys to perform visually guided stand and squat tasks. The current study, as a follow-up extension, investigates whether lower limb kinematics and muscle activity characterized by electromyography (EMG) signals during monkey performing stand/squat movements can be accurately decoded from neural spike trains in primary motor cortex (M1).

Two monkeys were used in this study. Subdermal intramuscular EMG electrodes were implanted to 8 right leg/thigh muscles. With ample data collected from neurons from a large brain area, we performed a spike triggered average (SpTA) analysis and got a series of density contours which revealed the spatial distributions of different muscle-innervating neurons corresponding to each given muscle. Based on the guidance of these results, we identified the locations optimal for chronic electrode implantation and subsequently carried on chronic neural data recordings. A recursive Bayesian estimation framework was proposed for decoding EMG signals together with kinematics from M1 spike trains.

Two specific algorithms were implemented: a standard Kalman filter and an unscented Kalman filter. For the latter one, an artificial neural network was incorporated to deal with the nonlinearity in neural tuning. High correlation coefficient and signal to noise ratio between the predicted and the actual data were achieved for both EMG signals and kinematics on both monkeys. Higher decoding accuracy and faster convergence rate could be achieved with the unscented Kalman filter. These results demonstrate that lower limb EMG signals and kinematics during monkey stand/squat can be accurately decoded from a group of M1 neurons with the proposed algorithms. Our findings provide new insights for extending current BMI design concepts and techniques on upper limbs to lower limb circumstances. Brain controlled exoskeleton, prostheses or neuromuscular electrical stimulators for lower limbs are expected to be developed, which enables the subject to manipulate complex biomechatronic devices with mind in more harmonized manner.

Background: Determining neuronal topographical information in the cerebral cortex is of fundamental importance for developing neuroprosthetics. Significant progress has been achieved in decoding hand voluntary movement with cortical neuronal activity in nonhuman primates. However, there are few successful reports in scientific literature for decoding lower limb voluntary movement with the cortical neuronal firing. We once reported an experimental system, which consists of a specially designed chair, a visually guided stand and squat task training paradigm and an acute neuron recording setup. With this system, we can record high quality cortical neuron activity to investigate the correlation between these neuronal signals and stand/squat movement.

Methods/Results: In this research, we train two monkeys to perform the visually guided stand and squat task, and record neuronal activity in the vast areas targeted to M1 hind-limb region, at a distance of 1 mm. We find that 76.9% of recorded neurons (1230 out of 1598 neurons) showing task-firing modulation, including 294 (18.4%) during the pre-response window; 310 (19.4%) for standing up; 104 (6.5%) for the holding stand phase; and 205 (12.8%) during the sitting down. The distributions of different type neurons have a high degree of overlap. They are mainly ranged from +7.0 to 13 mm in the Posterior-Anterior dimension, and from +0.5 to 4.0 mm in Dosal-lateral dimension, very close to the midline, and just anterior of the central sulcus.

Conclusions/Significance: The present study examines the neuronal activity related to lower limb voluntary movements in M1 and find topographical information of various neurons tuned to different stages of the stand and squat task. This work may contribute to understanding the fundamental principles of neural control of lower limb movements. Especially, the topographical information suggests us where to implant the chronic microelectrode arrays to harvest the most quantity and highest quality neurons related to lower limb movements, which may accelerate to develop cortically controlled lower limb neuroprosthetics for spinal cord injury subjects.

Millions of Americans live with motor impairments resulting from a stroke and the best way to administer rehabilitative therapy to achieve recovery is not well understood. Adaptive mixed reality rehabilitation (AMRR) is a novel integration of motion capture technology and high-level media computing that provides precise kinematic measurements and engaging multimodal feedback for self-assessment during a therapeutic task. The AMRR system was evaluated in a small (N=3) cohort of stroke survivors to determine best practices for administering adaptive, media-based therapy. A proof of concept study followed, examining changes in clinical scale and kinematic performances among a group of stroke survivors who received either a month of AMRR therapy (N = 11) or matched dosing of traditional repetitive task therapy (N = 10). Both groups demonstrated statistically significant improvements in Wolf Motor Function Test and upper-extremity Fugl-Meyer Assessment scores, indicating increased function after the therapy. However, only participants who received AMRR therapy showed a consistent improvement in their kinematic measurements, including those measured in the trained reaching task (reaching to grasp a cone) and in an untrained reaching task (reaching to push a lighted button). These results suggest that that the AMRR system can be used as a therapy tool to enhance both functionality and reaching kinematics that quantify movement quality. Additionally, the AMRR concepts are currently being transitioned to a home-based training application. An inexpensive, easy-to-use, toolkit of tangible objects has been developed to sense, assess and provide feedback on hand function during different functional activities. These objects have been shown to accurately and consistently track hand function in people with unimpaired movements and will be tested with stroke survivors in the future.

Multiple Sclerosis, an autoimmune disease, is one of the most common neurological disorder in which demyelinating of the axon occurs. The main symptoms of MS disease are fatigue, vision problems, stability issue, balance problems. Unfortunately, currently available treatments for this disease do not always guarantee the improvement of the condition of the MS patient and there has not been an accurate mechanism to measure the effectiveness of the treatment due to inter-patient heterogeneity. The factors that count for varying the performance of MS patients include environmental setting, weather, psychological status, dressing style and more. Also, patients may react differently while examined at specially arranged setting and this may not be the same while he/she is at home. Hence, it becomes a major problem for MS patients that how effectively a treatment slows down the progress of the disease and gives a relief for the patient. This thesis is trying to build a reliable system to estimate how good a treatment is for MS patients. Here I study the kinematic variables such as velocity of walking, stride length, variability and so on to find and compare the variations of the patient after a treatment given by the doctor, and trace these parameters for some patients after the treatment effect subdued.

An accurate sense of upper limb position is crucial to reaching movements where sensory information about upper limb position and target location is combined to specify critical features of the movement plan. This dissertation was dedicated to studying the mechanisms of how the brain estimates the limb position in space and the consequences of misestimation of limb position on movements. Two independent but related studies were performed. The first involved characterizing the neural mechanisms of limb position estimation in the non-human primate brain. Single unit recordings were obtained in area 5 of the posterior parietal cortex in order to examine the role of this area in estimating limb position based on visual and somatic signals (proprioceptive, efference copy). When examined individually, many area 5 neurons were tuned to the position of the limb in the workspace but very few neurons were modulated by visual feedback. At the population level however decoding of limb position was somewhat more accurate when visual feedback was provided. These findings support a role for area 5 in limb position estimation but also suggest that visual signals regarding limb position are only weakly represented in this area, and only at the population level. The second part of this dissertation focused on the consequences of misestimation of limb position for movement production. It is well known that limb movements are inherently variable. This variability could be the result of noise arising at one or more stages of movement production. Here we used biomechanical modeling and simulation techniques to characterize movement variability resulting from noise in estimating limb position ('sensing noise') and in planning required movement vectors ('planning noise'), and compared that to the variability expected due to noise in movement execution. We found that the effects of sensing and planning related noise on movement variability were dependent upon both the planned movement direction and the initial configuration of the arm and were different in many respects from the effects of execution noise.

Virtual environments are used for many physical rehabilitation and therapy purposes with varying degrees of success. An important feature for a therapy environment is the real-time monitoring of a participants' movement performance. Such monitoring can be used to evaluate the environment in addition to the participant's learning. Methods for monitoring and evaluation include tracking kinematic performance as well as monitoring muscle and brain activities through EMG and EEG technology. This study aims to observe trends in individual participants' motor learning based on changes in kinematic parameters and use those parameters to characterize different types of learners. This information can then guide EEG/EMG data analysis in the future. The evaluation of motor learning using kinematic parameters of performance typically compares averages of pre- and post-data to identify patterns of changes of various parameters. A key issue with using pre- and post-data is that individual participants perform differently and have different time-courses of learning. Furthermore, different parameters can evolve at independent rates. Finally, there is great variability in the movements at early stages of learning a task. To address these issues, a combined approach is proposed using robust regression, piece-wise regression and correlation to categorize different participant's motor learning. Using the mixed reality rehabilitation system developed at Arizona State University, it was possible to engage participants in motor learning, as revealed by improvements in kinematic parameters. A combination of robust regression, piecewise regression and correlation were used to reveal trends and characterize participants based on motor learning of three kinematic parameters: trajectory error, supination error and the number of phases in the velocity profile.