Neuron selection by relative importance for neural decoding of dexterous finger prosthesis control application

Future generations of upper limb prosthesis will have dexterous hand with individual fingers and will be controlled directly by neural signals. Neurons from the primary motor (M1) cortex code for finger movements and provide the source for neural control of dexterous prosthesis. Each neuron's activation can be quantified by the change in firing rate before and after finger movement, and the quantified value is then represented by the neural activity over each trial for the intended movement. Since this neural activity varies with the intended movement, we define the relative importance of each neuron independent of specific intended movements. The relative importance of each neuron is determined by the inter-movement variance of the neural activities for respective intended movements. Neurons are ranked by the relative importance and then a subpopulation of rank-ordered neurons is selected for the neural decoding. The use of the proposed neuron selection method in individual finger movements improved decoding accuracy by 21.5% in the case of decoding with only 5 neurons and by 9.2% in the case of decoding with only 10 neurons. With only 15 highly ranked neurons, a decoding accuracy of 99.5% was achieved. The performance improvement is still maintained when combined movements of two fingers were included though the decoding accuracy fell to 95.7%. Since the proposed neuron selection method can achieve the targeting accuracy of decoding algorithms with less number of input neurons, it can be significant for developing brain–machine interfaces for direct neural control of hand prostheses.     

fNIRS Exhibits Weak Tuning to Hand Movement Direction

Functional near-infrared spectroscopy (fNIRS) has become an established tool to investigate brain function and is, due to its portability and resistance to electromagnetic noise, an interesting modality for brain-machine interfaces (BMIs). BMIs have been successfully realized using the decoding of movement kinematics from intra-cortical recordings in monkey and human. Recently, it has been shown that hemodynamic brain responses as measured by fMRI are modulated by the direction of hand movements. However, quantitative data on the decoding of movement direction from hemodynamic responses is still lacking and it remains unclear whether this can be achieved with fNIRS, which records signals at a lower spatial resolution but with the advantage of being portable. Here, we recorded brain activity with fNIRS above different cortical areas while subjects performed hand movements in two different directions. We found that hemodynamic signals in contralateral sensorimotor areas vary with the direction of movements, though only weakly. Using these signals, movement direction could be inferred on a single-trial basis with an accuracy of ∼65% on average across subjects. The temporal evolution of decoding accuracy resembled that of typical hemodynamic responses observed in motor experiments. Simultaneous recordings with a head tracking system showed that head movements, at least up to some extent, do not influence the decoding of fNIRS signals. Due to the low accuracy, fNIRS is not a viable alternative for BMIs utilizing decoding of movement direction. However, due to its relative resistance to head movements, it is promising for studies investigating brain activity during motor experiments.     

Consistency of Listing’s law and reciprocal innervation with pseudo-inverse control of eye position in 3-D

Pseudo-inverse kinematics, under which small movements are produced by the least possible sum square changes in motor command, has been proposed as a unifying principle for the elimination of redundancy in general biological motor control systems (Pellionisz 1984) and in particular in the oculomotor system (Daunicht 1988, 1991). We have noted elsewhere (Dean et al. 1999) that this principle is incomplete without first specifying a parameterisation of motor command space and we proposed that the relevant motor command parameter is summed motor unit firing rate. Under this assumption we were able to show that pseudo-inverse control of the horizontal extraocular muscles is consistent with available motor pool firing rate data. In this paper we extend this result to three dimensions and all six extraocular muscles, showing that pseudo-inverse control is consistent with published firing rate data for a realistic model of oculomotor kinematics. We suggest that pseudo-inverse control may represent a common currency for modular control of many degree of freedom systems while its implementation may be a consequence of the minimisation of a more ecologically relevant parameter such as post-saccadic retinal slip.     

Recurrent cerebellar architecture solves the motor-error problem

Current views of cerebellar function have been heavily influenced by the models of Marr and Albus, who suggested that the climbing fibre input to the cerebellum acts as a teaching signal for motor learning. It is commonly assumed that this teaching signal must be motor error (the difference between actual and correct motor command), but this approach requires complex neural structures to estimate unobservable motor error from its observed sensory consequences. We have proposed elsewhere a recurrent decorrelation control architecture in which Marr-Albus models learn without requiring motor error. Here, we prove convergence for this architecture and demonstrate important advantages for the modular control of systems with multiple degrees of freedom. These results are illustrated by modelling adaptive plant compensation for the three-dimensional vestibular ocular reflex. This provides a functional role for recurrent cerebellar connectivity, which may be a generic anatomical feature of projections between regions of cerebral and cerebellar cortex.     

A collaborative brain-computer interface for improving human performance

Electroencephalogram (EEG) based brain-computer interfaces (BCI) have been studied since the 1970s. Currently, the main focus of BCI research lies on the clinical use, which aims to provide a new communication channel to patients with motor disabilities to improve their quality of life. However, the BCI technology can also be used to improve human performance for normal healthy users. Although this application has been proposed for a long time, little progress has been made in real-world practices due to technical limits of EEG. To overcome the bottleneck of low single-user BCI performance, this study proposes a collaborative paradigm to improve overall BCI performance by integrating information from multiple users. To test the feasibility of a collaborative BCI, this study quantitatively compares the classification accuracies of collaborative and single-user BCI applied to the EEG data collected from 20 subjects in a movement-planning experiment. This study also explores three different methods for fusing and analyzing EEG data from multiple subjects: (1) Event-related potentials (ERP) averaging, (2) Feature concatenating, and (3) Voting. In a demonstration system using the Voting method, the classification accuracy of predicting movement directions (reaching left vs. reaching right) was enhanced substantially from 66% to 80%, 88%, 93%, and 95% as the numbers of subjects increased from 1 to 5, 10, 15, and 20, respectively. Furthermore, the decision of reaching direction could be made around 100-250 ms earlier than the subject's actual motor response by decoding the ERP activities arising mainly from the posterior parietal cortex (PPC), which are related to the processing of visuomotor transmission. Taken together, these results suggest that a collaborative BCI can effectively fuse brain activities of a group of people to improve the overall performance of natural human behavior.     

Forest Walk Methods for Localizing Body Joints from Single Depth Image

We present multiple random forest methods for human pose estimation from single depth images that can operate in very high frame rate. We introduce four algorithms: random forest walk, greedy forest walk, random forest jumps, and greedy forest jumps. The proposed approaches can accurately infer the 3D positions of body joints without additional information such as temporal prior. A regression forest is trained to estimate the probability distribution to the direction or offset toward the particular joint, relative to the adjacent position. During pose estimation, the new position is chosen from a set of representative directions or offsets. The distribution for next position is found from traversing the regression tree from new position. The continual position sampling through 3D space will eventually produce an expectation of sample positions, which we estimate as the joint position. The experiments show that the accuracy is higher than current state-of-the-art pose estimation methods with additional advantage in computation time.     

Simplifying the estimation of diagnostic testing accuracy over time for high specificity tests in the absence of a gold standard

Many different methods for evaluating diagnostic test results in the absence of a gold standard have been proposed. In this paper, we discuss how one common method, a maximum likelihood estimate for a latent class model found via the Expectation-Maximization (EM) algorithm can be applied to longitudinal data where test sensitivity changes over time. We also propose two simplified and nonparametric methods which use data-based indicator variables for disease status and compare their accuracy to the maximum likelihood estimation (MLE) results. We find that with high specificity tests, the performance of simpler approximations may be just as high as the MLE.     

Motor unit number estimation--a Bayesian approach

All muscle contractions are dependent on the functioning of motor units. In diseases such as amyotrophic lateral sclerosis (ALS), progressive loss of motor units leads to gradual paralysis. A major difficulty in the search for a treatment for these diseases has been the lack of a reliable measure of disease progression. One possible measure would be an estimate of the number of surviving motor units. Despite over 30 years of motor unit number estimation (MUNE), all proposed methods have been met with practical and theoretical objections. Our aim is to develop a method of MUNE that overcomes these objections. We record the compound muscle action potential (CMAP) from a selected muscle in response to a graded electrical stimulation applied to the nerve. As the stimulus increases, the threshold of each motor unit is exceeded, and the size of the CMAP increases until a maximum response is obtained. However, the threshold potential required to excite an axon is not a precise value but fluctuates over a small range leading to probabilistic activation of motor units in response to a given stimulus. When the threshold ranges of motor units overlap, there may be alternation where the number of motor units that fire in response to the stimulus is variable. This means that increments in the value of the CMAP correspond to the firing of different combinations of motor units. At a fixed stimulus, variability in the CMAP, measured as variance, can be used to conduct MUNE using the "statistical" or the "Poisson" method. However, this method relies on the assumptions that the numbers of motor units that are firing probabilistically have the Poisson distribution and that all single motor unit action potentials (MUAP) have a fixed and identical size. These assumptions are not necessarily correct. We propose to develop a Bayesian statistical methodology to analyze electrophysiological data to provide an estimate of motor unit numbers. Our method of MUNE incorporates the variability of the threshold, the variability between and within single MUAPs, and baseline variability. Our model not only gives the most probable number of motor units but also provides information about both the population of units and individual units. We use Markov chain Monte Carlo to obtain information about the characteristics of individual motor units and about the population of motor units and the Bayesian information criterion for MUNE. We test our method of MUNE on three subjects. Our method provides a reproducible estimate for a patient with stable but severe ALS. In a serial study, we demonstrate a decline in the number of motor unit numbers with a patient with rapidly advancing disease. Finally, with our last patient, we show that our method has the capacity to estimate a larger number of motor units.     

Numerical discretization-based estimation methods for ordinary differential equation models via penalized spline smoothing with applications in biomedical research

Differential equations are extensively used for modeling dynamics of physical processes in many scientific fields such as engineering, physics, and biomedical sciences. Parameter estimation of differential equation models is a challenging problem because of high computational cost and high-dimensional parameter space. In this article, we propose a novel class of methods for estimating parameters in ordinary differential equation (ODE) models, which is motivated by HIV dynamics modeling. The new methods exploit the form of numerical discretization algorithms for an ODE solver to formulate estimating equations. First, a penalized-spline approach is employed to estimate the state variables and the estimated state variables are then plugged in a discretization formula of an ODE solver to obtain the ODE parameter estimates via a regression approach. We consider three different order of discretization methods, Euler's method, trapezoidal rule, and Runge-Kutta method. A higher-order numerical algorithm reduces numerical error in the approximation of the derivative, which produces a more accurate estimate, but its computational cost is higher. To balance the computational cost and estimation accuracy, we demonstrate, via simulation studies, that the trapezoidal discretization-based estimate is the best and is recommended for practical use. The asymptotic properties for the proposed numerical discretization-based estimators are established. Comparisons between the proposed methods and existing methods show a clear benefit of the proposed methods in regards to the trade-off between computational cost and estimation accuracy. We apply the proposed methods t an HIV study to further illustrate the usefulness of the proposed approaches.     

Development of an in vivo method for determining material properties of passive myocardium

Calculation of mechanical stresses and strains in the left ventricular (LV) myocardium by the finite element (FE) method relies on adequate knowledge of the material properties of myocardial tissue. In this paper, we present a model-based estimation procedure to characterize the stress-strain relationship in passive LV myocardium. A 3D FE model of the LV myocardium was used, which included morphological fiber and sheet structure and a nonlinear orthotropic constitutive law with different stiffness in the fiber, sheet, and sheet-normal directions. The estimation method was based on measured wall strains. We analyzed the method's ability to estimate the material parameters by generating a set of synthetic strain data by simulating the LV inflation phase with known material parameters. In this way we were able to verify the correctness of the solution and to analyze the effects of measurement and model error on the solution accuracy and stability. A sensitivity analysis was performed to investigate the observability of the material parameters and to determine which parameters to estimate. The results showed a high degree of coupling between the parameters governing the stiffness in each direction. Thus, only one parameter in each of the three directions was estimated. For the tested magnitudes of added noise and introduced model errors, the resulting estimated stress-strain characteristics in the fiber and sheet directions converged with good accuracy to the known relationship. The sheet-normal stress-strain relationship had a higher degree of uncertainty as more noise was added and model error was introduced.     

A Dendritic Mechanism for Decoding Traveling Waves: Principles and Applications to Motor Cortex

Traveling waves of neuronal oscillations have been observed in many cortical regions, including the motor and sensory cortex. Such waves are often modulated in a task-dependent fashion although their precise functional role remains a matter of debate. Here we conjecture that the cortex can utilize the direction and wavelength of traveling waves to encode information. We present a novel neural mechanism by which such information may be decoded by the spatial arrangement of receptors within the dendritic receptor field. In particular, we show how the density distributions of excitatory and inhibitory receptors can combine to act as a spatial filter of wave patterns. The proposed dendritic mechanism ensures that the neuron selectively responds to specific wave patterns, thus constituting a neural basis of pattern decoding. We validate this proposal in the descending motor system, where we model the large receptor fields of the pyramidal tract neurons — the principle outputs of the motor cortex — decoding motor commands encoded in the direction of traveling wave patterns in motor cortex. We use an existing model of field oscillations in motor cortex to investigate how the topology of the pyramidal cell receptor field acts to tune the cells responses to specific oscillatory wave patterns, even when those patterns are highly degraded. The model replicates key findings of the descending motor system during simple motor tasks, including variable interspike intervals and weak corticospinal coherence. By additionally showing how the nature of the wave patterns can be controlled by modulating the topology of local intra-cortical connections, we hence propose a novel integrated neuronal model of encoding and decoding motor commands.     

A quantitative study of command elements for abdominal positioning behavior in the crayfish Procambarus clarkii

We develop a statistical method to estimate the total number of command elements devoted to abdominal positioning behavior in crayfish. We assumed that all command elements can be identified, that each identified cell is equivalent to a tagged individual in a population, and that the cells were sampled randomly. Samples of 29, 30, 20, and 35 cells from abdominal ganglia A3, A4, A5, and A6, respectively, were taken from our catalog. We characterized each cell using several morphological and physiological criteria, determined how many times each identified cell was present in the sample, and estimated the total number of command elements using both a maximum likelihood method and a modification of the Lincoln index. The larger the proportion of identified cells seen only once in the sample, the more identified cells there were that were unrepresented in the sample. We estimate there are approximately 34, 60, 86, and 98 command elements in ganglia A3, A4, A5, and A6, respectively. Using a slightly different data set we show that the motor output of unipolar cells is more often stronger in the direction of the cell's axonal projection. In bipolar command elements, the output strength was uncorrelated with the relative sizes of the two projecting axons. No two cells in our sample were completely identical, and this morphological variability sometimes made it difficult to determine whether or not two cells obtained from different individuals were the same identified cell. We discuss why caution should be exercised in studies requiring precision in cell identification.     

Coordinates transformation and learning control for visually-guided voluntary movement with iteration: A Newton-like method in a function space

In order to control visually-guided voluntary movements, the central nervous system (CNS) must solve the following three computational problems at different levels: (1) determination of a desired trajectory in the visual coordinates, (2) transformation of the coordinates of the desired trajectory to the body coordinates and (3) generation of motor command. In this paper, the second and the third problems are treated at computational, representational and hardware levels of Marr. We first study the problems at the computational level, and then propose an iterative learning scheme as a possible algorithm. This is a trial and error type learning such as repetitive training of golf swing. The amount of motor command needed to coordinate activities of many muscles is not determined at once, but in a step-wise, trial and error fashion in the course of a set of repetitions. Actually, the motor command in the (n+1)-th iteration is a sum of the motor command in then-th iteration plus two modification terms which are, respectively, proportional to acceleration and speed errors between the desired trajectory and the realized trajectory in then-th iteration. We mathematically formulate this iterative learning control as a Newton-like method in functional spaces and prove its convergence under appropriate mathematical conditions with use of dynamical system theory and functional analysis. Computer simulations of this iterative learning control of a robotic manipulator in the body or visual coordinates are shown. Finally, we propose that areas 2, 5, and 7 of the sensory association cortex are possible sites of this learning control. Further we propose neural network model which acquires transformation matrices from acceleration or velocity to motor command, which are used in these schemes.     

High accuracy decoding of movement target direction in non-human primates based on common spatial patterns of local field potentials

Background

The current development of brain-machine interface technology is limited, among other factors, by concerns about the long-term stability of single- and multi-unit neural signals. In addition, the understanding of the relation between potentially more stable neural signals, such as local field potentials, and motor behavior is still in its early stages.

Methodology/Principal Findings

We tested the hypothesis that spatial correlation patterns of neural data can be used to decode movement target direction. In particular, we examined local field potentials (LFP), which are thought to be more stable over time than single unit activity (SUA). Using LFP recordings from chronically implanted electrodes in the dorsal premotor and primary motor cortex of non-human primates trained to make arm movements in different directions, we made the following observations: (i) it is possible to decode movement target direction with high fidelity from the spatial correlation patterns of neural activity in both primary motor (M1) and dorsal premotor cortex (PMd); (ii) the decoding accuracy of LFP was similar to the decoding accuracy obtained with the set of SUA recorded simultaneously; (iii) directional information varied with the LFP frequency sub-band, being greater in low (0.3–4 Hz) and high (48–200 Hz) frequency bands than in intermediate bands; (iv) the amount of directional information was similar in M1 and PMd; (v) reliable decoding was achieved well in advance of movement onset; and (vi) LFP were relatively stable over a period of one week.

Conclusions/Significance

The results demonstrate that the spatial correlation patterns of LFP signals can be used to decode movement target direction. This finding suggests that parameters of movement, such as target direction, have a stable spatial distribution within primary motor and dorsal premotor cortex, which may be used for brain-machine interfaces.     

Decoding Accuracy in Supplementary Motor Cortex Correlates with Perceptual Sensitivity to Tactile Roughness

Perceptual sensitivity to tactile roughness varies across individuals for the same degree of roughness. A number of neurophysiological studies have investigated the neural substrates of tactile roughness perception, but the neural processing underlying the strong individual differences in perceptual roughness sensitivity remains unknown. In this study, we explored the human brain activation patterns associated with the behavioral discriminability of surface texture roughness using functional magnetic resonance imaging (fMRI). First, a whole-brain searchlight multi-voxel pattern analysis (MVPA) was used to find brain regions from which we could decode roughness information. The searchlight MVPA revealed four brain regions showing significant decoding results: the supplementary motor area (SMA), contralateral postcentral gyrus (S1), and superior portion of the bilateral temporal pole (STP). Next, we evaluated the behavioral roughness discrimination sensitivity of each individual using the just-noticeable difference (JND) and correlated this with the decoding accuracy in each of the four regions. We found that only the SMA showed a significant correlation between neuronal decoding accuracy and JND across individuals; Participants with a smaller JND (i.e., better discrimination ability) exhibited higher decoding accuracy from their voxel response patterns in the SMA. Our findings suggest that multivariate voxel response patterns presented in the SMA represent individual perceptual sensitivity to tactile roughness and people with greater perceptual sensitivity to tactile roughness are likely to have more distinct neural representations of different roughness levels in their SMA.     

Systematic Biases in Human Heading Estimation

Heading estimation is vital to everyday navigation and locomotion. Despite extensive behavioral and physiological research on both visual and vestibular heading estimation over more than two decades, the accuracy of heading estimation has not yet been systematically evaluated. Therefore human visual and vestibular heading estimation was assessed in the horizontal plane using a motion platform and stereo visual display. Heading angle was overestimated during forward movements and underestimated during backward movements in response to both visual and vestibular stimuli, indicating an overall multimodal bias toward lateral directions. Lateral biases are consistent with the overrepresentation of lateral preferred directions observed in neural populations that carry visual and vestibular heading information, including MSTd and otolith afferent populations. Due to this overrepresentation, population vector decoding yields patterns of bias remarkably similar to those observed behaviorally. Lateral biases are inconsistent with standard Bayesian accounts which predict that estimates should be biased toward the most common straight forward heading direction. Nevertheless, lateral biases may be functionally relevant. They effectively constitute a perceptual scale expansion around straight ahead which could allow for more precise estimation and provide a high gain feedback signal to facilitate maintenance of straight-forward heading during everyday navigation and locomotion.     

Analysis of the effects of firing rate and synchronization on spike-triggered averaging of multidirectional motor unit torque

Spike-triggered averaging (STA) of muscle force transients has often been used to estimate motor unit contractile properties, using the discharge of a motor unit within the muscle as the triggering events. For motor units that exert torque about multiple degrees-of-freedom, STA has also been used to estimate motor unit pulling direction. It is well known that motor unit firing rate and weak synchronization of motor unit discharges with other motor units in the muscle can distort STA estimates of contractile properties, but the distortion of STA estimates of motor unit pulling direction has not been thoroughly evaluated. Here, we derive exact equations that predict that STA decouples firing rate and synchronization distortion when used to estimate motor unit pulling direction. We derive a framework for analyzing synchronization, consider whether the distortion due to synchronization can be removed from STA estimates of pulling direction, and show that there are distributions of motor unit pulling directions for which STA is insensitive to synchronization. We conclude that STA may give insight into how motoneuronal synchronization is organized with respect to motor unit pulling direction. Action Editor: David Terman     

Accurate and Reliable Cancer Classification Based on Probabilistic Inference of Pathway Activity

With the advent of high-throughput technologies for measuring genome-wide expression profiles, a large number of methods have been proposed for discovering diagnostic markers that can accurately discriminate between different classes of a disease. However, factors such as the small sample size of typical clinical data, the inherent noise in high-throughput measurements, and the heterogeneity across different samples, often make it difficult to find reliable gene markers. To overcome this problem, several studies have proposed the use of pathway-based markers, instead of individual gene markers, for building the classifier. Given a set of known pathways, these methods estimate the activity level of each pathway by summarizing the expression values of its member genes, and use the pathway activities for classification. It has been shown that pathway-based classifiers typically yield more reliable results compared to traditional gene-based classifiers. In this paper, we propose a new classification method based on probabilistic inference of pathway activities. For a given sample, we compute the log-likelihood ratio between different disease phenotypes based on the expression level of each gene. The activity of a given pathway is then inferred by combining the log-likelihood ratios of the constituent genes. We apply the proposed method to the classification of breast cancer metastasis, and show that it achieves higher accuracy and identifies more reproducible pathway markers compared to several existing pathway activity inference methods.     

MODEM: a multi-agent hierarchical structure to model the human motor control system

In this study, based on behavioral and neurophysiological facts, a new hierarchical multi-agent architecture is proposed to model the human motor control system. Performance of the proposed structure is investigated by simulating the control of sit to stand movement. To develop the model, concepts of mixture of experts, modular structure, and some aspects of equilibrium point hypothesis were brought together. We have called this architecture MODularized Experts Model (MODEM). Human motor system is modeled at the joint torque level and the role of the muscles has been embedded in the function of the joint compliance characteristics. The input to the motor system, i.e., the central command, is the reciprocal command. At the lower level, there are several experts to generate the central command to control the task according to the details of the movement. The number of experts depends on the task to be performed. At the higher level, a “gate selector” block selects the suitable subordinate expert considering the context of the task. Each expert consists of a main controller and a predictor as well as several auxiliary modules. The main controller of an expert learns to control the performance of a given task by generating appropriate central commands under given conditions and/or constraints. The auxiliary modules of this expert learn to scrutinize the generated central command by the main controller. Auxiliary modules increase their intervention to correct the central command if the movement error is increased due to an external disturbance. Each auxiliary module acts autonomously and can be interpreted as an agent. Each agent is responsible for one joint and, therefore, the number of the agents of each expert is equal to the number of joints. Our results indicate that this architecture is robust against external disturbances, signal-dependent noise in sensory information, and changes in the environment. We also discuss the neurophysiological and behavioral basis of the proposed model (MODEM).