Adaptive tracking of EEG oscillations

https://doi.org/10.1016/j.jneumeth.2009.10.018Get rights and content

Abstract

Neuronal oscillations are an important aspect of EEG recordings. These oscillations are supposed to be involved in several cognitive mechanisms. For instance, oscillatory activity is considered a key component for the top-down control of perception. However, measuring this activity and its influence requires precise extraction of frequency components. This processing is not straightforward. Particularly, difficulties with extracting oscillations arise due to their time-varying characteristics. Moreover, when phase information is needed, it is of the utmost importance to extract narrow-band signals. This paper presents a novel method using adaptive filters for tracking and extracting these time-varying oscillations. This scheme is designed to maximize the oscillatory behavior at the output of the adaptive filter. It is then capable of tracking an oscillation and describing its temporal evolution even during low amplitude time segments. Moreover, this method can be extended in order to track several oscillations simultaneously and to use multiple signals. These two extensions are particularly relevant in the framework of EEG data processing, where oscillations are active at the same time in different frequency bands and signals are recorded with multiple sensors. The presented tracking scheme is first tested with synthetic signals in order to highlight its capabilities. Then it is applied to data recorded during a visual shape discrimination experiment for assessing its usefulness during EEG processing and in detecting functionally relevant changes. This method is an interesting additional processing step for providing alternative information compared to classical time–frequency analyses and for improving the detection and analysis of cross-frequency couplings.

Introduction

Oscillatory phenomena have been the focus of increasing interest in neuroscience research. Neuronal oscillations have been proposed as a key mechanism for the large-scale integration of cognitive processes through which top-down internal states influence stimulus processing (Engel et al., 2001, Varela et al., 2001). Several models have been developed, with oscillations either serving as a binding mechanism bringing together different perceptions into a unified representation (Singer and Gray, 1995, Engel and Singer, 2001) or as a dynamic substrate for neuronal communication achieved through the coherence between brain areas (Fries, 2005). Also a more precise observation of specific oscillatory parameters can shed light on even more detailed brain processes. For instance, the ongoing oscillatory state of the brain before a given stimulus has been shown to provide valuable information about the subsequent behavioral responses in both motor and sensory tasks (Linkenkaer-Hansen et al., 2004, Womelsdorf et al., 2006). Additionally, the phase of neuronal oscillations was successfully linked to activity of single neurons (Jacobs et al., 2007). Finally, increasing evidence indicates that responses within classical neuronal frequency bands likely interact with each other through coupling mechanisms that remain to be identified (Jensen and Colgin, 2007). In this framework, cross-frequency couplings could provide a unifying mechanism for the intermingled neuronal oscillations acting at different temporal and spatial scales (Von Stein and Sarnthein, 2000), and recent studies tend to verify the existence, and the possible importance of cross-frequency couplings, during a variety of motor, sensory and cognitive tasks (Canolty et al., 2006, Lakatos et al., 2007, Demiralp et al., 2007).

Taken together, these findings raise the need for efficient methods for accurate estimation of oscillatory information such as phase, frequency and amplitude from raw signals. A well-known method widely used to get such spectral information is the Hilbert transform and its analytic signal representation (Gabor, 1946). However, although many studies have successfully identified and described phase synchronizations by applying this method to wide-band neuronal signals, it has been shown that proper estimation of oscillatory parameters can be performed only on narrow-band signals (Nho and Loughlin, 1999, Chavez et al., 2006). Moreover, subsequent synchronization measures such as the Phase Locking Value (Tass et al., 1998) are reliable only when applied to narrow-band signals (Celka, 2007). Therefore, band-pass filtering was applied to neuronal signals as a pre-processing step, in order to split the raw signals into narrow-band oscillations of different frequencies. Although this filter bank approach can lead to a more reliable analysis of oscillatory interactions (Canolty et al., 2006), a major drawback of such pre-processing should be mentioned. Because the cut-off frequencies of each band-pass filter must be pre-defined and remain constant during the whole analysis window, physiologically misleading outputs could be produced by the filters, in the case of a frequency component crossing the cut-off frequency limit of a filter. In such situations, it would be preferable to follow an oscillatory component in a continuous manner, without constraining the spectral content to fixed limits. This remark emphasizes the need for adaptive methods able to track narrow-band oscillations over time.

Recently, we proposed a novel method for adaptively tracking multiple oscillations in single-trial EEG signals (Uldry et al., 2009). In this article, we describe the tracking abilities of our algorithm for the estimation of single or multiple frequencies in both synthetic and EEG signals. The physiological relevance of well-known synchronization measures can be assessed using the temporal outputs of our method. Importantly, our previous publication on this tracking scheme is extended in order to clearly illustrate its capabilities for adaptive frequency estimation and its advantages over more traditional approaches for measuring cross-frequency couplings. In Section 2, we present the basics of our algorithm as well as its multi-frequency and multi-signal extensions, and we illustrate its performance on synthetic signals. In Section 3, we present the results of our method on real EEG single-trial signals in terms of adaptive frequency tracking, and demonstrate the benefit of applying common synchronization measures on the temporal outputs of our filters, compared to current methods.

Section snippets

Methods

The oscillation tracking methods are presented within the complex-valued signal framework. This approach simplifies several aspects of the computations. Especially, the filters are shorter (only one pole is needed for a complex band-pass filter, whereas two poles are required for a real band-pass filter). Of course, the signals of interest are real-valued in practice. But with the Hilbert transform one obtains the so-called analytic representation, whose real part is the original signal itself.

Results

The data we present have been taken from a previously published study examining the spatio-temporal mechanisms of illusory contour perception with broad-band event-related potentials (Murray et al., 2002). This paradigm was chosen here as it represents a typical situation where time–frequency analyses based on wavelet decomposition have been extensively applied and have led to propositions regarding the role of gamma oscillations as a binding mechanism in human cortex (Tallon-Baudry and

Discussion

Oscillatory phenomena have gained increasing importance in the field of neuroscience, particularly because improvements in analysis methods have revealed how oscillatory activity is both a highly efficient and also information-rich signal. One paradigmatic shift in the conceptualization of oscillatory activity has been to consider not only changes within a particular frequency band, but also the interactions and synchronizations between frequencies of brain activity that are in turn thought to

Acknowledgments

We thank the anonymous reviewers for helpful and constructive comments. This work was supported by Swiss National Science Foundation grant 320030-120579. The Cartool software (http://brainmapping.unige.ch/Cartool.htm) has been programmed by Denis Brunet, from the Functional Brain Mapping Laboratory, Geneva, Switzerland, and is supported by the Center for Biomedical Imaging (CIBM) of Geneva and Lausanne.

References (40)

  • Y. Prudat et al.

    Multi-signal extension of adaptive frequency tracking algorithms

    Signal Process

    (2009)
  • C.E. Schroeder et al.

    Low-frequency neuronal oscillations as instruments of sensory selection

    Trends Neurosci

    (2009)
  • C. Tallon-Baudry et al.

    Oscillatory gamma activity in humans and its role in object representation

    Trends Cogn Sci

    (1999)
  • A. Von Stein et al.

    Different frequencies for different scales of cortical integration: from local gamma to long range alpha/theta synchronization

    Int J Psychphysiol

    (2000)
  • G. Buzsaki

    Rythms of the brain

    (2006)
  • R.T. Canolty et al.

    High-gamma power is phase locked to theta oscillations in human neocortex

    Science

    (2006)
  • P. Celka

    Statistical analysis of the phase-locking value

    IEEE Signal Process Lett

    (2007)
  • A.K. Engel et al.

    Dynamic predictions: oscillations and synchrony in top-down processing

    Nature

    (2001)
  • P. Flandrin

    Time–frequency/time-scale analysis

    (1999)
  • D. Gabor

    Theory of communication

    J IEE Lond

    (1946)
  • Cited by (34)

    • Detecting slow narrowband modulation in EEG signals

      2022, Journal of Neuroscience Methods
      Citation Excerpt :

      Power can also vary widely across individuals. Similarly, modulatory phenomena in the EEG can be highly diverse, with different frequencies being involved and implicated with various aspects of brain function Jensen and Colgin (2007); Van Zaen et al. (2010); Zandvoort and Nolte (2021). It remains unclear whether these phenomena are mechanistically related at the level of their underlying neural circuit dynamics.

    • Respiratory rate estimation from the ECG using an instantaneous frequency tracking algorithm

      2014, Biomedical Signal Processing and Control
      Citation Excerpt :

      This multi-signal frequency tracking method operates recursively on several signals simultaneously to track a common frequency component. This method has been shown to successfully track a common frequency component in biomedical signals such as electroencephalogram signals [9]. Furthermore, it is instantaneous and provides an automatic approach to RR estimation from the RSA and RPA, in contrast to the ad hoc processing proposed in [6].

    • Adaptive frequency tracking of the baseline ECG identifies the site of atrial fibrillation termination by catheter ablation

      2013, Biomedical Signal Processing and Control
      Citation Excerpt :

      We chose to circumvent this limitation by using adaptive bandpass filters. The harmonic frequency tracking algorithm presented in this article is an extension of a scheme proposed previously [20,21]. The basic tracking method is briefly introduced before the extension for harmonic components is described.

    • Illusory contours: A window onto the neurophysiology of constructing perception

      2013, Trends in Cognitive Sciences
      Citation Excerpt :

      ERPs can likewise be acquired simultaneously with transcranial magnetic stimulation (TMS) [24] or functional magnetic resonance imaging (fMRI) [25], allowing inferences to be drawn on the correspondence between ERP modulations and hemodynamic signals [26]. Decomposition of EEG or MEG into time–frequency representations allows assessment of the contribution of frequency and phase information to IC processing specifically and the binding problem more generally [27–32]. Independent series of studies by a number of laboratories have focused on the question of when and where IC sensitivity is first manifest (Figure 3).

    • PERCEPTION OF OWN AND ANOTHER FACE: MULTIANALYSIS APPROACH OF EEG DATA

      2022, Zhurnal Vysshei Nervnoi Deyatelnosti Imeni I.P. Pavlova
    View all citing articles on Scopus
    View full text