Nonstationarity in epileptic EEG and implications for neural dynamics

Manuca, Radu; Casdagli, Martin; Savit, Robert

doi:10.1016/s0025-5564(97)00055-2

Cited by 31 publications

(12 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…5 1). Further developments, including statistical tests that can distinguish between linear and nonlinear deterministic dynamics (e.g., 39,7 1,72), techniques that can identity subtle spatio-temporal interdependencies of the epileptogenic disturbance (40,73), and techniques that can take into consideration the highly nonstationary character of EEG time series (74,75) in real time. Therefore, nonlinear time series analyses are expected to contribute further to improved presurgical evaluation and anticipation of seizure onset, which might lead to the development of appropriate seizure prevention strategies in the near future.…”

Section: Perspectivesmentioning

confidence: 99%

Nonlinear EEG Analysis and Its Potential Role in Epileptology

et al. 2000

View full text Add to dashboard Cite

Summary: Deterministic chaos offers a striking explanation for apparently irregular behavior of the brain that is evidenced in the EEG. Recent developments in the physical-mathematical framework of the theory of nonlinear dynamics (colloquially often termed chaos theory) provide new concepts and powerful algorithms to analyze such time series. Because of its high versatility, nonlinear time series analysis has already gone beyond the physical sciences and, at present, is being successfully applied in a variety of disciplines, including cardiology, neurology, psychiatry, and epileptology. However, it is well known that different influencing factors limit the use of nonlinear measures to characterize EEG dynamics in a strict sense. Nevertheless, when interpreted with care, relative estimates of, e.g., Despite rapid advances in neuroimaging techniques (see ref. 1 for an overview), EEG recordings continue to play an important role in diagnosis of epilepsy. To extract relevant information from long-term EEG recordings, a variety of computerized analysis methods have been developed ( 2 4 ) . Most methods are based primarily on the assumption that EEG is generated by a highly complex linear system, resulting in characteristic signal features such as nonstationarity or unpredictability (5-7). On the other hand, EEG signals can be interpreted as the output of a deterministic system of relatively simple complexity but containing highly nonlinear elements. It has been shown that nonlinear dynamical systems, with at least three degrees of freedom, might exhibit chaotic behavior, thus becoming unpredictable over a long time scale (8). The assumption of reduced complexity is even more valid in the case of neurons or neuronal networks that participate in the epileptogenic process, taking into account elementary mechanisms underlying epilepsy @,lo). Various investigations have shown that applying nonlinear time series analysis (NTSA) to recordings of brain electrical activity offers new information about the complex dynamics of underlying neuronal networks (see refs. 11-1 3 for an overview). Within this physical-mathematical framework, a variety of measures allow characterization of different static and dynamical properties of a time series (14,15). The Lyapunov exponents (16,17) provide a measure for the degree of chaoticity and the correlation dimension (1 8) describes the number of degrees of freedom of the underlying dynamics. The latter and the Kolmogorov entropy (19) estimate the degree of ordeddisorder and thus of the complexity of a dynamic system. In a strict sense, well-known problems in extracting nonlinear measures from short, noisy, and nonstationary data would exclude the use of these measures for characterization of EEG dynamics. However, if interpreted with care, e.g., only relative measures with respect to time and recording site are assumed reliable, results of various investigations now provide converging evidence that information supplied by NTSA is superior to conventional parametric or nonparametric analys...

show abstract

Section: Perspectivesmentioning

confidence: 99%

Nonlinear EEG Analysis and Its Potential Role in Epileptology

et al. 2000

View full text Add to dashboard Cite

show abstract

“…The patient is seizure-free approximately from timepoints 0-3,500 (pre-ictal period), in a transition state from 3,501-6,000 and in seizure from 6,001-10,000 (ictal-period). The prediction errors using linear and nonlinear kernels were estimated locally by analyzing using time segments of length (T) 50 (number of samples) due to the non stationarity properties of EEG channels during epileptic seizures (Andrzejak et al, 2001;Manuca et al, 1998;Rankine et al, 2007). The SVR parameters and lag order were set identically to those used for the simulations.…”

Section: Application To An Eeg Datasetmentioning

confidence: 99%

Measuring Time Series Predictability Using Support Vector Regression

Sato

Costafreda

Morettin

et al. 2008

Communications in Statistics - Simulation and Computation

View full text Add to dashboard Cite

Most studies involving statistical time series analysis rely on assumptions of linearity, which by its simplicity facilitates parameter interpretation and estimation. However, the linearity assumption may be too restrictive for many practical applications. The implementation of nonlinear models in time series analysis involves the estimation of a large set of parameters, frequently leading to overfitting problems. In this article, a predictability coefficient is estimated using a combination of nonlinear autoregressive models and the use of support vector regression in this model is explored. We illustrate the usefulness and interpretability of results by using electroencephalographic records of an epileptic patient.

show abstract

“…On the other hand, EEG data has been shown to exhibit non-stationary behavior in a variety of contexts [35] and epileptiform EEG contains non-stationarity [41]. Specifically, some seizures had time intervals of a few seconds that were stationary according to several statistical criteria [67], even though the whole ictal event behaves as non-stationary phenomena [11,34,50,54,57]; in particular, the epileptic seizures can be divided into sequential 3-4 stages using dynamical non-stationary analysis which captures regime changes faster than the spectral and statistical analysis [12].…”

Section: Introductionmentioning

confidence: 99%