Showing papers by "Emery N. Brown published in 2007"
TL;DR: This work presents a novel method for global interference reduction and real-time recovery of evoked brain activity, based on the combination of a multiseparation probe configuration and adaptive filtering, and demonstrates that the physiological interference in the superficial layers is the major component of global interference.
Abstract: The sensitivity of near-infrared spectroscopy (NIRS) to evoked brain activity is reduced by physiological interference in at least two locations: 1. the superficial scalp and skull layers, and 2. in brain tissue itself. These interferences are generally termed as "global interferences" or "systemic interferences," and arise from cardiac activity, respiration, and other homeostatic processes. We present a novel method for global interference reduction and real-time recovery of evoked brain activity, based on the combination of a multiseparation probe configuration and adaptive filtering. Monte Carlo simulations demonstrate that this method can be effective in reducing the global interference and recovering otherwise obscured evoked brain activity. We also demonstrate that the physiological interference in the superficial layers is the major component of global interference. Thus, a measurement of superficial layer hemodynamics (e.g., using a short source-detector separation) makes a good reference in adaptive interference cancellation. The adaptive-filtering-based algorithm is shown to be resistant to errors in source-detector position information as well as to errors in the differential pathlength factor (DPF). The technique can be performed in real time, an important feature required for applications such as brain activity localization, biofeedback, and potential neuroprosthetic devices.
150 citations
01 Jul 2007
TL;DR: In this paper, a multiseparation probe configuration and adaptive filtering was proposed to reduce the global interference and recovering otherwise obscured evoked brain activity in near-infrared spectroscopy.
Abstract: The sensitivity of near-infrared spectroscopy (NIRS) to evoked brain activity is reduced by physiological interference in at least two locations: 1. the superficial scalp and skull layers, and 2. in brain tissue itself. These interferences are generally termed as "global interferences" or "systemic interferences," and arise from cardiac activity, respiration, and other homeostatic processes. We present a novel method for global interference reduction and real-time recovery of evoked brain activity, based on the combination of a multiseparation probe configuration and adaptive filtering. Monte Carlo simulations demonstrate that this method can be effective in reducing the global interference and recovering otherwise obscured evoked brain activity. We also demonstrate that the physiological interference in the superficial layers is the major component of global interference. Thus, a measurement of superficial layer hemodynamics (e.g., using a short source-detector separation) makes a good reference in adaptive interference cancellation. The adaptive-filtering-based algorithm is shown to be resistant to errors in source-detector position information as well as to errors in the differential pathlength factor (DPF). The technique can be performed in real time, an important feature required for applications such as brain activity localization, biofeedback, and potential neuroprosthetic devices.
137 citations
TL;DR: The results showed that the hemodynamic changes in the superficial layers and the estimated total global interference in the target measurement were highly correlated, which explains why this global interference cancellation method should work well when global interference is dominating.
Abstract: Following previous Monte Carlo simulations, we describe in detail an example of detecting evoked visual hemodynamic re- sponses in a human subject as a preliminary demonstration of the novel global interference cancellation technology. The raw time series of oxyhemoglobin O2Hb and deoxyhemoglobin HHb changes, their block averaged results before and after adaptive filtering, to- gether with the power spectral density analysis are presented. Simul- taneous respiration and EKG recordings suggested that spontaneous low-frequency oscillation and cardiac activity were the major sources of global interference in our test. When global interference dominates e.g., for O2Hb in our data, judged by power spectral density analy- sis, adaptive filtering effectively reduced this interference, doubling the contrast-to-noise ratio CNR for evoked visual response detection. When global interference is not obvious e.g., in our HHb data, adaptive filtering provided no CNR improvement. The results also showed that the hemodynamic changes in the superficial layers and the estimated total global interference in the target measurement were highly correlated r=0.96, which explains why this global interfer- ence cancellation method should work well when global interference is dominating. In addition, the results suggested that association be- tween the superficial layer hemodynamics and the total global inter- ference is time-varying. © 2007 Society of Photo-Optical Instrumentation Engineers. DOI: 10.1117/1.2804706
118 citations
TL;DR: This model provides a novel framework to explore dynamical principles that may underlie normal and pathologic sleep-wake physiology and suggests distinct network mechanisms for the two types of wakefulness.
Abstract: Recent work in experimental neurophysiology has identified distinct neuronal populations in the rodent brain stem and hypothalamus that selectively promote wake and sleep. Mutual inhibition between these cell groups has suggested the conceptual model of a sleep-wake switch that controls transitions between wake and sleep while minimizing time spent in intermediate states. By combining wake- and sleep-active populations with populations governing transitions between different stages of sleep, a "sleep-wake network" of neuronal populations may be defined. To better understand the dynamics inherent in this network, we created a model sleep-wake network composed of coupled relaxation oscillation equations. Mathematical analysis of the deterministic model provides insight into the dynamics underlying state transitions and predicts mechanisms for each transition type. With the addition of noise, the simulated sleep-wake behavior generated by the model reproduces many qualitative and quantitative features of mouse sleep-wake behavior. In particular, the existence of simulated brief awakenings is a unique feature of the model. In addition to capturing the experimentally observed qualitative difference between brief and sustained wake bouts, the model suggests distinct network mechanisms for the two types of wakefulness. Because circadian and other factors alter the fine architecture of sleep-wake behavior, this model provides a novel framework to explore dynamical principles that may underlie normal and pathologic sleep-wake physiology.
111 citations
01 Nov 2007
TL;DR: In this paper, an example of detecting evoked visual hemodynamic responses in a human subject was presented as a preliminary demonstration of the novel global interference cancellation technology. But adaptive filtering provided no CNR improvement.
Abstract: Following previous Monte Carlo simulations, we describe in detail an example of detecting evoked visual hemodynamic re- sponses in a human subject as a preliminary demonstration of the novel global interference cancellation technology. The raw time series of oxyhemoglobin O2Hb and deoxyhemoglobin HHb changes, their block averaged results before and after adaptive filtering, to- gether with the power spectral density analysis are presented. Simul- taneous respiration and EKG recordings suggested that spontaneous low-frequency oscillation and cardiac activity were the major sources of global interference in our test. When global interference dominates e.g., for O2Hb in our data, judged by power spectral density analy- sis, adaptive filtering effectively reduced this interference, doubling the contrast-to-noise ratio CNR for evoked visual response detection. When global interference is not obvious e.g., in our HHb data, adaptive filtering provided no CNR improvement. The results also showed that the hemodynamic changes in the superficial layers and the estimated total global interference in the target measurement were highly correlated r=0.96, which explains why this global interfer- ence cancellation method should work well when global interference is dominating. In addition, the results suggested that association be- tween the superficial layer hemodynamics and the total global inter- ference is time-varying. © 2007 Society of Photo-Optical Instrumentation Engineers. DOI: 10.1117/1.2804706
107 citations
TL;DR: A coherent estimation framework is reported that outperforms previous approaches under various conditions, in the control of position and velocity, based on trajectory and endpoint mean squared errors.
Abstract: Brain-driven interfaces depend on estimation procedures to convert neural signals to inputs for prosthetic devices that can assist individuals with severe motor deficits. Previous estimation procedures were developed on an application-specific basis. Here we report a coherent estimation framework that unifies these procedures and motivates new applications of prosthetic devices driven by action potentials, local field potentials (LFPs), electrocorticography (ECoG), electroencephalography (EEG), electromyography (EMG), or optical methods. The brain-driven interface is described as a probabilistic relationship between neural activity and components of a prosthetic device that may take on discrete or continuous values. A new estimation procedure is developed for action potentials, and a corresponding procedure is described for field potentials and optical measurements. We test our framework against dominant approaches in an arm reaching task using simulated traces of ensemble spiking activity from primary motor cortex (MI) and a wheelchair navigation task using simulated traces of EEG-band power. Adaptive filtering is incorporated to demonstrate performance under neuron death and discovery. Finally, we characterize performance under model misspecification using physiologically realistic history dependence in MI spiking. These simulated results predict that the unified framework outperforms previous approaches under various conditions, in the control of position and velocity, based on trajectory and endpoint mean squared errors.
96 citations
TL;DR: Two new SMC point process filters are constructed using sequential Monte Carlo approximations to the BCK equations and provide more accurate state estimates at low number of particles than a conventional bootstrap SMC filter algorithm in which the state transition probability density is the proposal density.
Abstract: The stochastic state point process filter (SSPPF) and steepest descent point process filter (SDPPF) are adaptive filter algorithms for state estimation from point process observations that have been used to track neural receptive field plasticity and to decode the representations of biological signals in ensemble neural spiking activity. The SSPPF and SDPPF are constructed using, respectively, Gaussian and steepest descent approximations to the standard Bayes and Chapman-Kolmogorov (BCK) system of filter equations. To extend these approaches for constructing point process adaptive filters, we develop sequential Monte Carlo (SMC) approximations to the BCK equations in which the SSPPF and SDPPF serve as the proposal densities. We term the two new SMC point process filters SMC-PPFS and SMC-PPFD , respectively. We illustrate the new filter algorithms by decoding the wind stimulus magnitude from simulated neural spiking activity in the cricket cercal system. The SMC-PPFS and SMC-PPFD provide more accurate state estimates at low number of particles than a conventional bootstrap SMC filter algorithm in which the state transition probability density is the proposal density. We also use the SMC-PPFS algorithm to track the temporal evolution of a spatial receptive field of a rat hippocampal neuron recorded while the animal foraged in an open environment. Our results suggest an approach for constructing point process adaptive filters using SMC methods
83 citations
01 Jan 2007
TL;DR: The stochastic state point process filter (SSPPF) and steepest descent point process filtering (SDPPF) as mentioned in this paper are adaptive filter algorithms for state estimation from point process observations that have been used to track neural receptive field plasticity and decode the representations of biological signals in ensemble neural spiking activity.
Abstract: The stochastic state point process filter (SSPPF) and steepest descent point process filter (SDPPF) are adaptive filter algorithms for state estimation from point process observations that have been used to track neural receptive field plasticity and to decode the representations of biological signals in ensemble neural spiking activity. The SSPPF and SDPPF are constructed using, respectively, Gaussian and steepest descent approximations to the standard Bayes and Chapman-Kolmogorov (BCK) system of filter equations. To extend these approaches for constructing point process adaptive filters, we develop sequential Monte Carlo (SMC) approximations to the BCK equations in which the SSPPF and SDPPF serve as the proposal densities. We term the two new SMC point process filters - and - ,r e- spectively. We illustrate the new filter algorithms by decoding the wind stimulus magnitude from simulated neural spiking activity in the cricket cercal system. The - and - provide more accurate state estimates at low number of particles than a conventional bootstrap SMC filter algorithm in which the state transition probability density is the proposal density. We also use the - algorithm to track the temporal evolution of a spatial receptive field of a rat hippocampal neuron recorded while the animal foraged in an open environment. Our results suggest an approach for constructing point process adaptive filters using SMC methods.
74 citations
TL;DR: A Bayesian analysis of a state-space model for characterizing simultaneous learning of multiple tasks and for assessing behavioral biases in learning experiments with interleaved task presentations is presented and an improved, computationally efficient approach is suggested.
Abstract: Accurate characterizations of behavior during learning experiments are essential for understanding the neural bases of learning. Whereas learning experiments often give subjects multiple tasks to l...
54 citations
TL;DR: A framework to combine sensor data and control algorithms along with neural activity and state equations, to coordinate goal-directed movements through brain-driven interfaces is proposed.
Abstract: State-space estimation is a convenient framework for the design of brain-driven interfaces, where neural activity is used to control assistive devices for individuals with severe motor deficits. Recently, state-space approaches were developed to combine goal planning and trajectory-guiding neural activity in the control of reaching movements of an assistive device to static goals. In this paper, we extend these algorithms to allow for goals that may change over the course of the reach. Performance between static and dynamic goal state equations and a standard free movement state equation is compared in simulation. Simulated trials are also used to explore the possibility of incorporating activity from parietal areas that have previously been associated with dynamic goal position. Performance is quantified using mean-square error (MSE) of trajectory estimates. We also demonstrate the use of goal estimate MSE in evaluating algorithms for the control of goal-directed movements. Finally, we propose a framework to combine sensor data and control algorithms along with neural activity and state equations, to coordinate goal-directed movements through brain-driven interfaces
28 citations
12 Apr 2007
TL;DR: The expectation-maximization (EM) algorithm is applied to estimate parameters and sources in an MEG state-space model and it is demonstrated in simulation studies that the resulting source estimates are superior to those provided by static methods or dynamic methods employing ad hoc parameter selection.
Abstract: Dynamic estimation methods based on linear state-space models have been applied to the inverse problem of magnetoencephalography (MEG), and can improve source localization compared with static methods by incorporating temporal continuity as a constraint. The efficacy of these methods is influenced by how well the state-space model approximates the dynamics of the underlying brain current sources. While some components of the state-space model can be inferred from brain anatomy and knowledge of the MEG instrument noise structure, parameters governing the temporal evolution of underlying current sources are unknown and must be selected on an ad-hoc basis or estimated from data. In this work, we apply the expectation-maximization (EM) algorithm to estimate parameters and sources in an MEG state-space model and demonstrate in simulation studies that the resulting source estimates are superior to those provided by static methods or dynamic methods employing ad hoc parameter selection.
02 May 2007
TL;DR: It is demonstrated that inter-subject variability in brain response to an exercise task may help explain the natural variability in autonomic response, as assessed by HRV analysis.
Abstract: While the central autonomic network (CAN) has been adequately defined in animal models, data from the human have been lacking. In this study, we correlated cardiac-gated fMRI data with continuous-time heart rate variability (HRV) assessment in order to estimate central autonomic processing in response to a dynamic grip task. The electrocardiogram (ECG) was collected simultaneously with fMRI, and was analyzed with a new point process adaptive filter algorithm for evaluation of HRV indices reflecting time-varying dynamics of autonomic modulation. These were correlated with fMRI signal intensity using a general linear model and subsequent analysis of covariance. Our combined HRV-fMRI data analysis suggests that fMRI activity in several brain regions, including the hypothalamus, parabrachial nucleus, periaqueductal gray, amygdala, and posterior insula, demonstrated significant correlation with parasympathetic tone assessed by HRV high frequency (HF) power. This study demonstrates that inter-subject variability in brain response to an exercise task may help explain the natural variability in autonomic response, as assessed by HRV analysis
01 Dec 2007
TL;DR: A state-space generalized linear model for characterizing neural spiking activity in multiple trials (SS-GLM) is presented and the model is used to quantify the neural changes related to learning in hippocampal neural activity recorded from a monkey performing a location-scene association task.
Abstract: We present a state-space generalized linear model (SS-GLM) for characterizing neural spiking activity in multiple trials. We estimate the model parameters by maximum likelihood using an approximate Expectation-maximization (EM) algorithm which employs a recursive point process filter, fixed-interval smoothing and state-space covariance algorithms. We assess model goodness-of-fit using the time-rescaling theorem and guide the choice of model order with Akaike's information criterion. We illustrate our approach in two applications. In the analysis of hippocampal neural activity recorded from a monkey performing a location-scene association task, we use the model to quantify the neural changes related to learning. In the analysis of primary auditory cortex responses to different levels of electrical stimulation in the rat midbrain, we use the method to analyze auditory threshold detection. Our findings have important implications for developing theoretically-sound and practical tools to characterize the dynamics of spiking activity.