Simulated generation of evoked potentials components using networkswith distinct excitatory and inhibitory neurons

Long latency evoked potentials (EPs) are electrical potentials related to brain information processing mechanisms. A three-layered neurophysiologically based artificial neural network model is presented whose neurons obey to Dale's law. The first two layers of the network can memorize and recall sparsely coded patterns, oscillating at biologically plausible frequencies. Excitatory low-pass filtering synapses, from the second to the third layer, create evoked current dipoles, when the network retrieves memories related to stimuli. Based on psychophysiological indications, simulated intracranial dipoles are straightforwardly transformed into long latency EP components such as N ₁₀₀ and P₃₀₀ that match laboratory-measured scalp EPs     

A robust parametric estimator for single-trial movement relatedbrain potentials

Current estimators for single-trial evoked potentials (EP's) require a signal-to-noise ratio (SNR) of 0 dB or better to obtain high quality estimations, yet many types of EP's suffer from substantially lower SNR's. This paper presents a robust-evoked-potential-estimator (REPE) facilitating high quality estimations of single movement related EP's with a relatively low SNR. The estimator is based on a standard ARX model, enhanced to support estimation under poor SNR conditions. The REPE was tested successfully on a computer simulated data set giving reliable single-trial estimations for the low SNR range of around -20 dB. The REPE was also applied to experimental data, producing clear single-trial estimations of movement related brain signals recorded in a classic scenario of self-paced finger tapping experiment     

Nonlinear alignment and averaging for estimating the evokedpotential

Addresses the problems associated with averaging brain responses evoked through a repetitive application of an external stimulus. In order to improve the estimate of the evoked potential (EP) through signal averaging, a method which incorporates nonlinear alignment of the EPs into the averaging operation is developed. The method makes no prior assumptions about the properties of the EP or which response in the set best characterizes the EP to be estimated. The nonlinear alignment procedure is designed to pairwise generate optimally aligned EPs by backtracking along the optimal alignment path. The nonlinear alignment and averaging operations are systematically combined to develop methods to estimate the EP. Results from a series of experiments conducted on simulated and real sets of responses show that, through nonlinear alignment and averaging, the events in the EPs are preserved and the estimates of the EP are quite robust     

Single trial evoked potential analysis by means of crosscorrelation and dynamic time-warping

A new approach to obtain evoked potential (EP) estimates is described. EPs are minute voltage fluctuations in the electrical activity of the brain in response to an external stimulus. EPs are often buried in the spontaneous brain activity and special techniques are required to improve the signal-to-noise ratio. The method presented here uses a dynamic time-warping (DTW) algorithm to maximize the crosscorrelation between a template and the recorded block. The algorithm was evaluated using simulated data consisting of triangular shapes superimposed on white noise and a rhythm resembling alpha activity (8–12 Hz).It was found that warping increased the crosscorrelation between the template and single trial, however, the template occasionally converged to the background noise especially for low signal-to-noise ratios. For SNRs larger than 2, the time-warping approach obtained better latency estimates than the conventional averaging method. Amplitude estimates were consistently inferior for DTW.     

Evaluation and application of a RBF neural network for online single-sweep extraction of SEPs during scoliosis surgery

A method for on-line single sweep detection of somatosensory evoked potentials (SEPs) during intraoperative neuromonitoring is proposed. It is based on a radial-basis function neural network with Gaussian activations. In order to improve its tracking capabilities, the radial-basis functions location is partially learnt sweep-by-sweep; the training algorithm is effective, though consistent with real-time applications. This new detection method has been tested on simulated data so as to set the network parameters. Moreover, it has been applied to real recordings obtained from a new neuromonitoring technique which is based on the simultaneous observation of the SEP and of the evoked H-reflex elicited by the same electric stimulus. The SEPs have been extracted using the neural network and the results have then been compared to those obtained by ARX filtering and correlated with the spinal cord integrity information obtained by the H-reflex. The proposed algorithm has been proved to be particularly effective and suitable for single-sweep detection. It is able to track both sudden and smooth signal changes of both amplitude and latency and the needed computational time is moderate.     

Multichannel fusion models for the parametric classification of differential brain activity

This paper introduces parametric multichannel fusion models to exploit the different but complementary brain activity information recorded from multiple channels in order to accurately classify differential brain activity into their respective categories. A parametric weighted decision fusion model and two parametric weighted data fusion models are introduced for the classification of averaged multichannel evoked potentials (EPs). The decision fusion model combines the independent decisions of each channel classifier into a decision fusion vector and a parametric classifier is designed to determine the EP class from the discrete decision fusion vector. The data fusion models include the weighted EP-sum model in which the fusion vector is a linear combination of the multichannel EPs and the EP-concatenation model in which the fusion vector is a vector-concatenation of the multichannel EPs. The discrete Karhunen-Loeve transform (DKLT) is used to select features for each channel classifier and from each data fusion vector. The difficulty in estimating the probability density function (PDF) parameters from a small number of averaged EPs is identified and the class conditional PDFs of the feature vectors of averaged EPs are, therefore, derived in terms of the PDFs of the single-trial EPs. Multivariate parametric classifiers are developed for each fusion strategy and the performances of the different strategies are compared by classifying 14-channel EPs collected from five subjects involved in making explicit match/mismatch comparisons between sequentially presented stimuli. It is shown that the performance improves by incorporating weights in the fusion rules and that the best performance is obtained using multichannel EP concatenation. It is also noted that the fusion strategies introduced are also applicable to other problems involving the classification of multicategory multivariate signals generated from multiple sources.     

Development and evaluation of the piecewise Prony method for evoked potential analysis

A new method is presented to decompose nonstationary signals into a summation of oscillatory components with time varying frequency, amplitude, and phase characteristics. This method, referred to as piecewise Prony method (PPM), is an improvement over the classical Prony method, which can only deal with signals containing components with fixed frequency, amplitude and phase, and monotonically increasing or decreasing rate of change. PPM allows the study of the temporal profile of post-stimulus signal changes in single-trial evoked potentials (EPs), which can lead to new insights in EP generation. We have evaluated this method on simulated data to test its limitations and capabilities, and also on single-trial EPs. The simulation experiments showed that the PPM can detect amplitude changes as small as 10%, rate changes as small as 10%, and 0.15 Hz of frequency changes. The capabilities of the PPM were demonstrated using single electroencephalogram/EP trials of flash visual EPs recorded from one normal subject. The trial-by-trial results confirmed that the stimulation drastically attenuates the alpha activity shortly after stimulus presentation, with the alpha activity returning about 0.5 s later. The PPM results also provided evidence that delta activity undergoes phase alignment following stimulus presentation.     

Estimation of single-evoked cerebral potentials by means ofparametric modeling and Kalman filtering

A method is presented for the investigation of stimulus evoked cerebral potentials (EPs) in single-trial EEG recordings which separates the measured activity into its evoked and spontaneous parts. A compound state-space model trying to incorporate the observable properties of both parts is formulated on the basis of additivity of the two components. Within this model, spontaneous activity is described as an autoregressive process, while the EP is modeled by an impulse response of a parametrically described system. Based on the state-space representation of the model, a Kalman filter for the observation of the system's state can be designed which yields optimal estimates for both activities. The properties of the proposed method are tested by application to stimulated data, in which preset EPs are added to measured spontaneous EEG segments. The relative mean squared error and the bias are used to judge the accuracy of the EPs retrieved by the filter. Finally, the method is applied to data where rapid drug-induced effects can be monitored with high time-resolution by means of estimated somatosensory evoked potentials     

Real-time data-reusing adaptive learning of a radial basis function network for tracking evoked potentials

Tracking variations in both the latency and amplitude of evoked potential (EP) is important in quantifying properties of the nervous system. Adaptive filtering is a powerful tool for tracking such variations. In this paper, a data-reusing non-linear adaptive filtering method, based on a radial basis function network (RBFN), is implemented to estimate EP. The RBFN consists of an input layer of source nodes, a single hidden layer of non-linear processing units and an output layer of linear weights. It has built-in nonlinear activation functions that allow learning of function mappings. Moreover, it produces satisfactory estimates of signals against a background noise without a priori knowledge of the signal, provided that the signal and noise are independent. In clinical situations where EP responses change rapidly, the convergence rate of the algorithm becomes a critical factor. A carefully designed data-reusing RBFN can accelerate the convergence rate markedly and, thus, enhance its performance. Both theoretical analysis and simulation results support the improved performance of our new algorithm.     

诱发脑电提取方法的研究进展

诱发脑电(EP)的提取是生物医学信号处理领域中的一个热点研究方向,针对诱发脑电提取的现状,阐述了相干平均、独立分量分析、小波变换、时间序列分析及神经网络等方法的基本原理,并探讨了各自在应用中存在的问题。其为后续研究提供了理论基础。     

Noise reduction in brain evoked potentials based on third-order correlations

In this paper, we use third-order correlations (TOC) in developing a filtering technique for the recovery of brain evoked potentials (EPs). The main idea behind the presented technique is to pass the noisy signal through a finite impulse response filter whose impulse response is matched with the shape of the noise-free signal. It is shown that it is possible to estimate the filter impulse response on basis of a selected third-order correlation slice (TOCS) of the input noisy signal. This is justified by two facts. The first one is that the noise-free EPs can be modeled as a sum of damped sinusoidal signals and the selected TOCS preserve the signal structure. The second fact is that the TOCS is insensitive to both Gaussian noise and other symmetrically distributed non-Gaussian noise, (white or colored). Furthermore, the approach can be applied to either nonaveraged or averaged EP observation data. In the nonaveraged data case, the approach therefore preserves information about amplitude and latency changes. Both fixed and adaptive versions of the proposed filtering technique are described. Extensive simulation results are provided to show the validity and effectiveness of the proposed cumulant-based filtering technique in comparison with the conventional correlation-based counterpart.     

Detection of evoked potentials with the use of the stochastic filtering theory

For detecting evoked potentials (EPs) in electroencephalogram (EEG) signals, a mathematical model of EEG without EP is constructed and equations of stochastic filtering are synthesized. From the analysis of updating sequences, an EP detection criterion is formulated. An EP detection algorithm is developed and studied, and its efficiency is confirmed by means of simulation.     

联合稀疏表示的双次诱发电位提取算法

诱发电位少次提取对于研究大脑活动规律以及临床诊断等具有重要意义.根据脑电信号的特点,本文提出一种基于联合稀疏表示的双次诱发电位信号估计算法.利用诱发电位信号的准周期性和自发脑电信号的随机性,该算法将脑电信号看作为相似成分和相异成分的叠加.神经系统通过相同刺激产生的诱发电位主要在潜伏期和波幅两方面发生变化,因此该算法利用平均诱发电位进行建模,得到稀疏字典,通过联合稀疏表示算法实现双次诱发电位信号的提取.实验结果表明,该算法和其他算法相比获得了更好的效果.     

The Detection of Auditory Evoked Responses Using a Matched Filter

A method has been developed to detect auditory brainstem evoked responses (ABR's) using a minimum amount of computer averaging. The method employs a matched filter to detect the ABR buried in the EEG. The matched filter system can also be used to predict wave V latency, which is useful in testing for hearing loss. By using a matched filter derived from an ABR obtained at a high-stimulus level, it is possible to calculate wave V latency at lower intensity levels much faster. Excellent correlation is seen between wave V latencies calculated in this manner, and those obtained after significant amounts of averaging. Because of this, the matched filter system can reduce the amount of time required for a hearing loss test from 20?30 min to approximately 5?10 min with no significant degradation in results.     

Microwave-evoked brainstem potentials in cats.

Rectangular pulse-modulated microwave radiation has been shown to produce auditory responses in mammals. It is therefore reasonable to explore the possibility of using microwave pulses to achieve an estimate of sensori-neural involvement in the objective evaluation of human hearing and to assess the presence of tumors or brainstem lesions in patients with neurological disorders. In this paper we shall show that microwave-evoked auditory response of cats closely resembles that evoked by acoustic pulse. We shall also give preliminary results obtained from electrodes fastened to the vertex of the skull after successive coagulative production of lesions in the inferior colliculus, lateral lemniscus, and superior olivary nucleus.     

A method for real-time processing to study recovery functions ofevoked potentials

A method for evaluating the recovery function of the evoked potential (EP) in real time is reported. To record the test response of paired stimulation, the responses to alternate presentation of an unpaired stimulus and several kinds of paired stimuli with various interstimulus intervals were averaged, and subtraction was made in real time. The EPs obtained by the proposed method were proved to be accurate even under changing recording conditions. A formula for computing the recovery correlation factor was derived as a measure of the recovery function of the recorded EP, and the recovery curves for three subjects were obtained     

Adaptive estimation of latency changes in evoked potentials

Changes in latency of evoked potentials (EP) may indicate clinically and diagnostically important changes in the status of the nervous system. A low signal-to-noise ratio of the EP signal makes it difficult to estimate small, transient, time-varying changes in latency, or delays. Here, the authors present an adaptive algorithm that estimates small delay (latency change) values even when EP signal amplitudes are time-varying. When the delay is time invariant, the adaptive algorithm produces an unbiased estimate with delay estimation error less than half of the sampling interval. A lower estimation error variance is obtained when, in a pair of signals, the adaptive algorithm delays the signal with the higher SNR. The adaptive delay estimation algorithm was tested on intra-operative recordings of somatosensory EP, and analysis of those recordings reveals that the anesthetic etomidate produces a step change in the amplitude and latency of the EP signals     

Nonlinear system identification by m-pulse sequences:application to brainstem auditory evoked responses

The purpose of this paper is to introduce a method for characterizing the nonlinear behavior of the auditory system. The method uses an m-pulse sequence as the stimulus and employs a general nonlinear framework for the auditory system. Like Sutter's binary m-sequence approach, the m-pulse sequence approach is computationally efficient since calculation of the first-order input-output cross-correlation function is all that is necessary for obtaining the nonlinear characteristics of the system. The nonlinear system characteristics are reflected in pulse kernels in contrast to binary kernels associated with the binary m-sequence approach. By assuming the system under study is a third-order nonlinear system, binary and pulse kernels are shown to be related to Volterra kernels. The results suggest that the m-pulse sequence can be used to study the system nonlinear effects of varying the stimulus repetition rate more effectively than conventional methods. Preliminary physiological data obtained by applying m-pulse sequences to the brainstem auditory evoked response (BAER) clearly illustrates the feasibility of obtaining replicable evoked responses using this method.     

A Maximum Likelihood Method for Estimating EEG Evoked Potentials

The problem of EEG evoked potential (EP) estimation is basically one of estimating a transient signal embedded in nonstationary mostly additive noise; and as such it is well suited to a nonstationary estimation approach utilizing, for example, the Kalman filter. The method presented in this paper is based on a model of the EEG response which is assumed to be the sum of the EP and independent correlated Gaussian noise representing the spontaneous EEG activity. The EP is assumed to vary in both shape and latency; the latter is assumed to be governed by some unspecified probability density; and no assumption on stationarity is needed for the noise. With the model described in state-space form, a Kalman filter is constructed, and the variance of the innovation process is derived; a maximum likelihood solution to the EP estimation problem is then obtained via this innovation process. The method was tested on simulated as well as real EEG data.     

MEG covariance difference analysis: a method to extract targetsource activities by using task and control measurements

A method is proposed for extracting target dipole-source activities from two sets of evoked magnetoencephalographic (MEG) data, one measured using task stimuli and the other using control stimuli. The difference matrix between the two covariance matrices obtained from these two measurements is calculated, and a procedure similar to the MEG-multiple signal classification (MUSIC) algorithm is applied to this difference matrix to extract the target dipole-source configuration. This configuration corresponds to the source-configuration difference between the two measurements. Computer simulation verified the validity of the proposed method. The method was applied to actual evoked-field data obtained from simulated task-and-control experiments. In these measurements, a combination of auditory and somatosensory stimuli was used as the task stimulus and the somatosensory stimulus alone was used as the control stimulus. The proposed covariance difference analysis successfully extracted the target auditory source and eliminated the disturbance from the somatosensory sources