Spike detection and sorting is a fundamental step in the analysis of extracellular neural recording. Here, we propose a combined spike detection-feature extraction algorithm that relies on a sparse representation space of the spike waveforms. The proposed method captures the wavelet footprint of the waveform, by calculating the power of the scale space vectors and finding an optimal detection threshold using histogram equalization techniques. Under the proposed scheme, a compact feature set is obtained simultaneously during detection, which eliminates the need for a separate feature extraction step for spike sorting. Our results demonstrate that this method yields improved performance, particularly in low SNRs, while preserving the separability between neuronal clusters in the feature space.

This paper presents the implementation of a low-power and implantable neuroprocessor on low-cost nano-FPGA for data reduction and on-the-fly spike sorting in Brain Machine Interface applications. Detailed analysis of efficient utilization of the hardware resources, power consumption and design scalability are provided. The prototype we report here enables simultaneous processing of 32-channel data sampled at 25 kHz/channel with 8-bit/sample resolution with less than 5 mW power consumption for all modes of operation (monitoring, compression and sensing) at 1.2 V core voltage supply on a 5 mm × 5 mm nano-FPGA.

Multi-channel neural signal recordings need high data compression and efficient data transmission. Our previous work has shown a practical data compression solution based on discrete wavelet transform, multi-level thresholding and run length encoding. This paper presents a custom designed communication protocol for bidirectional data telemetry to and from the implanted module. A global controller is also presented which configures, operates and unites all the modules together effectively and efficiently into a 32-channel system. Performance of the communication protocol and the compression engine is analyzed.

Coordination among cortical neurons is believed to be a key element in mediating many high-level cortical processes such as perception, attention, learning, and memory formation. Inferring the structure of the neural circuitry underlying this coordination is important to characterize the highly nonlinear, time-varying interactions between cortical neurons in the presence of complex stimuli. In this work, we investigate the applicability of dynamic Bayesian networks (DBNs) in inferring the effective connectivity between spiking cortical neurons from their observed spike trains. We demonstrate that DBNs can infer the underlying nonlinear and time-varying causal interactions between these neurons and can discriminate between mono- and polysynaptic links between them under certain constraints governing their putative connectivity. We analyzed conditionally Poisson spike train data mimicking spiking activity of cortical networks of small and moderately large size. The performance was assessed and compared to other methods under systematic variations of the network structure to mimic a wide range of responses typically observed in the cortex. Results demonstrate the utility of DBN in inferring the effective connectivity in cortical networks.

An essential step towards understanding how the brain orchestrates information processing at the cellular and population levels is to simultaneously observe the spiking activity of cortical neurons that mediate perception, learning, and motor processing. In this paper, we formulate an information theoretic approach to determine whether cooperation among neurons may constitute a governing mechanism of information processing when encoding external covariates. Specifically, we show that conditional independence between neuronal outputs may not provide an optimal encoding strategy when the firing probability of a neuron depends on the history of firing of other neurons connected to it. Rather, cooperation among neurons can provide a "message-passing" mechanism that preserves most of the information in the covariates under specific constraints governing their connectivity structure. Using a biologically plausible statistical learning model, we demonstrate the performance of the proposed approach in synergistically encoding a motor task using a subset of neurons drawn randomly from a large population. We demonstrate its superiority in approximating the joint density of the population from limited data compared to a statistically independent model and a maximum entropy (MaxEnt) model.

Estimating rate functions underlying neural point processes is essential for characterizing the firing patterns of cortical neurons involved in sensory and motor processing. This paper introduces a new method for directly estimating neuronal firing rates from a compressed representation of their extracellular recordings. The approach is based on extending a near-optimal sparse representation of the extracellular recordings to time scales matching those of the underlying rate functions, thereby performing the same role as a kernel-density estimator but at a much lower computational cost. Experimental results demonstrate that this method achieves comparable performance to the conventional Gaussian kernel estimation methods at a fraction of the computational cost. This makes it suitable to brain machine interface applications where real time decoding of neural activity for neuro-motor prosthetic control using firing rate estimates is strongly desired.

A fundamental goal in systems neuroscience is to infer the functional connectivity among neuronal elements coordinating information processing in the brain. In this work, we investigate the applicability of dynamic Bayesian networks (DBN) in inferring the structure of cortical networks from the observed spike trains. DBNs have unique features that make them capable of detecting causal relationships between spike trains such as modeling time-dependent relationships, detecting non-linear interactions and inferring connectivity between neurons from the observed ensemble activity. A probabilistic point process model was used to assess the performance under systematic variations of the model parameters. Results demonstrate the utility of DBN in inferring functional connectivity in cortical network models.

Neural recordings from high-density microelectrode arrays implanted in the cortex require time-frequency domain processing to alleviate the data telemetry bottlenecks of bandwidth and power. Our previous work has shown that the energy compaction capability of the discrete wavelet transform (DWT) offers a practical data compression solution that faithfully preserves the information in the neural signals. This paper presents a complete compression system including both lossy and lossless compression schemes, namely the DWT and run length encoding. Performance tradeoffs and key design decisions for implantable applications are analyzed. A 32-channel, 4-level version of the circuit is presented. Custom designed in 0.5 mum CMOS, occupying only 5.75 mm2and consuming 3mW of power (95 muW per channel at 25Ks/sec), the implantable compression circuit is well suited for intra-cortical neural interface applications.

Understanding the mechanism underlying distributed neural coding is a fundamental goal in computational neuroscience. With the ability to simultaneously observe the activity of large networks of neurons in response to external stimuli, a natural question arises: how the outside world is represented in the collective activity of these neurons? In this work, we provide an information theoretic approach for determining the role of cooperation among neurons in encoding external stimuli. Specifically, we show that statistical independence between neuronal outputs may not provide the best coding strategy when these outputs depend on the history of other neuronal constituents in the network. Rather, cooperation among neurons can provide a near optimal and lossless coding strategy under specific constraints governing their network structure. Using a statistical learning model, we demonstrate the performance of the proposed approach in decoding a motor task with both discrete targets and continuous trajectory using spike trains from a small subset of a large network. We demonstrate its superiority in minimizing the decoding error compared to a statistically independent model and to other classical decoders reported in the literature.

Identifying functional connectivity between neuronal elements is an essential first step toward understanding how the brain orchestrates information processing at the single-cell and population levels to carry out biological computations. This letter suggests a new approach to identify functional connectivity between neuronal elements from their simultaneously recorded spike trains. In particular, we identify clusters of neurons that exhibit functional interdependency over variable spatial and temporal patterns of interaction. We represent neurons as objects in a graph and connect them using arbitrarily defined similarity measures calculated across multiple timescales. We then use a probabilistic spectral clustering algorithm to cluster the neurons in the graph by solving a minimum graph cut optimization problem. Using point process theory to model population activity, we demonstrate the robustness of the approach in tracking a broad spectrum of neuronal interaction, from synchrony to rate co-modulation, by systematically varying the length of the firing history interval and the strength of the connecting synapses that govern the discharge pattern of each neuron. We also demonstrate how activity-dependent plasticity can be tracked and quantified in multiple network topologies built to mimic distinct behavioral contexts. We compare the performance to classical approaches to illustrate the substantial gain in performance.

Brain machine interfaces (BMIs) have recently received significant attention from the neuroscience and engineering communities as a result of striking advances in monitoring, processing, and modeling brain function at multiple temporal and spatial resolutions. These advances, however, have also raised significant challenges to both communities that are becoming the focus of numerous ongoing research efforts. Broadly categorized based on their level of invasiveness, BMIs relying on implantable microelectrode arrays (MEAs) have received the most attention. This paper briefly reviews some fundamental concepts underlying the operation of MEA-based BMIs and highlights in particular the signal processing challenges faced by these systems in light of their resource-constrained operation. Finally, we summarize some of our recent progress in this area and suggest some open questions for future research.

This paper proposes a new approach to simultaneously estimate time-varying intensity functions of multiple point processes from their continuous-time signal representation. We use models of neural response properties in the cortex to illustrate the theory of the proposed approach. Based on sparse representation of the continuous-time signals in the context of compression, it is shown that intensity functions can be approximated reasonably well without the need to decompress and classify the source signals. The approach is best suited for the case when multiple point processes are characterized by non-binary spike waveforms observed with an array of sensors. When spike waveforms from different sources are correlated, the estimated intensities can be inaccurate due to spike classification errors. We therefore build on our previous work for separating correlated spike waveforms to enable enhanced separation of those intensity functions. We finally show that this framework leads to substantial savings in computational complexity for real time operation in resource constrained signal processing systems.

Discrete wavelet transform (DWT) has been shown to provide exceptionally efficient data compression for neural records. This paper describes an area-power minimized hardware implementation of the lifting scheme for multi-level, multi-channel DWT. Performance tradeoffs and key design decisions for implantable neuroprosthetics are analyzed. A 32-channel, 4-level version of the circuit has been custom designed in 0.18μm CMOS and occupies only 0.16mm2.

Identifying clusters of neurons with correlated spiking activity in large-size neuronal ensembles recorded with high-density multielectrode array is an emerging problem in computational neuroscience. We propose a nonparametric approach that represents multiple neural spike trains by a mixed point process model. A spectral clustering algorithm is applied to identify the clusters of neurons through their correlated firing activities. The advantage of the proposed technique is its ability to efficiently identify large populations of neurons with correlated spiking activity independent of the temporal scale. We report the clustering performance of the algorithm applied to a complex synthesized data set and compare it to multiple clustering techniques.

We investigate a new approach for the problem of source separation/classification of non-orthogonal, non-stationary noisy signals impinging on an array of sensors. We propose a solution to the problem when the contaminating noise is temporally and spatially correlated. The observations are projected onto a nested set of multiresolution spaces prior to classical eigendecomposition. An inherent invariance property of the signal subspace is observed in a subset of the multiresolution spaces that depends on the level of approximation expressed by the orthogonal basis. This feature, among others revealed by the algorithm, is eventually used to separate the correlated signal sources in the context of 'best basis' selection. The technique shows robustness to source non-stationarity as well as anisotropic properties of the channel characteristics under no constraints on the array design. We illustrate the high performance of the technique on simulated and experimental multichannel neurophysiological data measurements

We conceptualize a new signal processing strategy to better represent the temporal and spectral cues in speech signals for hearing aid (HA) and cochlear implant (CI) applications under severe adverse conditions. The proposed approach rests on two well studied methods for signal separation and noise suppression, namely, the denoising and function approximation capabilities of the wavelet transform, blended with signal subspace decomposition through low rank approximation. The technique targets suppression of "competing voice" type noises. A cost function is defined to obtain a "best basis" representation of the desired speech signal for which an inherent invariance property of the signal subspace is observed. This allows better separation of the speech-like noise in contrast to classical bandpass filtering currently employed in CI and HA devices. We demonstrate the efficiency of the proposed method in capturing the rapid dynamics of speech signals, while minimizing the masking effects of noise, in addition to improved recognition rates in normal hearing listeners. The technique remains to be tested on actual patients

Data fusion from closely spaced nonuniformly distributed sensor arrays with large sensor count requires a smart mechanism for minimizing information redundancy while maintaining high signal fidelity. In this work, we investigate the detection of multiple signal sources impinging on a nonuniformly spaced array of sensors when the additive noise has local spatial correlation characteristics. In particular, we show that if the signal spatial correlation length does not match the noise correlation length, significant performance loss occurs because of improper subarray selection. We propose to formulate the log-likelihood function (LLF) in the multiresolution domain by spatially whitening the wavelet-transformed observations followed by local universal thresholding. The LLF obtained is independent of the noise covariance term, thereby yielding significant improvement over classical time domain LLF tests. Performance comparison with data averaging and decision fusion detectors is carried out to illustrate the potential advantages of the proposed method.

The problem of detecting the number of sources impinging on an array of sensors has received wide interest in many research problems. In particular, the detection of the number of distinct neural sources using a recording array of closely spaced sensors in the brain is one such application. The special case of transient source signals of unknown waveforms corrupted by Gaussian noise is the focus of this paper. We propose a new approach for solving this problem when no a priori knowledge is given about the neural sources and/or the noise processes. By extending our previous array multiresolution analysis framework for noise suppression, signal detection and identification (Oweiss and Anderson (2000, 2001, 2002)), we show that it is feasible to achieve reasonable source detection performance in moderate to low SNR scenarios.

The problem of detecting transient signals of unknown waveforms has been widely studied in recent years due to the numerous applications associated with it. In particular, the detection of neural spikes using a recording array of closely spaced sensors in the brain is one such application. In this paper, we propose a new approach for solving this problem when no apriori knowledge is given about the signal and/or the noise processes. By extending our previous multiresolution analysis framework for noise suppression and source identification [1,2], we show that it is feasible to achieve reasonable detection performance in low SNR scenarios. Comparison to traditional detection schemes is presented and the overall performance is evaluated.

Technology for recording action potentials from the nervous system has changed in recent years from single or at most two or three neurons to tens of neurons recorded from geometrically precise arrays of recording sites. To fully exploit these research opportunities, signal delivery and processing are key factors. We desire not only to sort the array input into neural channels but to locate neurons with respect to the array so they can be later identified in histology and serve to further decode the activity of each neuron. The distribution of a neuron's signal across the sensor array is predictive of the cell location at least in projection onto the array but also in the dimension above the array. To achieve 3-dimensional locating power, careful estimation of signal strength at each site must be achieved taking into account the distance and the field distortion for all anticipated source positions. Achieving this independent of source strength appears to be feasible.

We investigated a new framework for the problem of blind source identification in multichannel signal processing. Inspired by a neurophysiological data environment, where an array of closely spaced recording electrodes is surrounded by multiple neural cell sources, significant spatial correlation of source signals motivated the need for an efficient technique for reliable multichannel blind source identification. In a previous work Oweiss and Anderson (see Proceedings of the 34th. Asilomar Conference on Signals, Systems and Computers, Pacific Grove, 2000) adopted a new approach for noise suppression based on thresholding an array discrete wavelet transform (ADWT) representation of the multichannel data. We extend the work of Oweiss and Anderson to identify sources from the observation mixtures. The technique relies on separating sources with highest spatial energy distribution in each frequency subband spanned by the corresponding wavelet basis. Accordingly, the best basis selection criterion we propose benefits from the additional degree of freedom offered by the space domain. The amplitude and shift invariance properties revealed by this technique make it very efficient to track spatial source variations sometimes encountered in multichannel neural recordings. Results from multichannel multiunit neural data are presented and the overall performance is evaluated.