The measure of randomness by leave-one-out prediction error in the analysis of EEG after laser painful stimulation in healthy subjects and migraine patients

TIRES-Center of Innovative Technologies for Signal Detection and Processing, University of Bari, Italy.
Clinical Neurophysiology (Impact Factor: 3.1). 01/2006; 116(12):2775-82. DOI: 10.1016/j.clinph.2005.08.019
Source: PubMed


We aimed to perform a quantitative analysis of event-related modulation of EEG activity, resulting from a not-warned and a warned paradigm of painful laser stimulation, in migraine patients and controls, by the use of a novel analysis, based upon a parametric approach to measure predictability of short and noisy time series.
Ten migraine patients were evaluated during the not-symptomatic phase and compared to seven age and sex matched controls. The dorsum of the right hand and the right supraorbital zone were stimulated by a painful CO(2) laser, in presence or in absence of a visual warning stimulus. An analysis time of 1s after the stimulus was submitted to a time-frequency analysis by a complex Morlet wavelet and to a cross-correlation analysis, in order to detect the development of EEG changes and the most activated cortical regions. A parametric approach to measure predictability of short and noisy time series was applied, where time series were modeled by leave-one-out (LOO) error.
The averaged laser-evoked potentials features were similar between the two groups in the alerted and not alerted condition. A strong reset of the beta rhythms after the painful stimuli was seen for three groups of electrodes along the midline in patients and controls: the predictability of the series induced by the laser stimulus changed very differently in controls and patients. The separation was more evident after the warning signal, leading to a separation with P-values of 0.0046 for both the hand and the face.
As painful stimulus causes organization of the local activity in cortex, EEG series become more predictable after stimulation. This phenomenon was less evident in migraine, as a sign of an inadequate cortical reactivity to pain.
The LOO method enabled to show in migraine subtle changes in the cortical response to pain.

  • No preview · Article · Jan 2006 · Clinical Neurophysiology
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Spontaneous EEG patterns are studied to detect migraine patients both during the attack and in headache-free periods. The EEG signals are analyzed through the wavelets and both scale-dependent and scale-independent features are computed to characterize the patterns. The classification is carried out by a supervised neural network. The efficiency of the method is evaluated through the Receiver Operating Characteristic (ROC) analysis and the Wilcoxon-Mann-Whitney (WMW) test. Although a high discrimination is observed with one single neural output, a complete separation among MwA patients and healthy subjects is obtained when a scatter plot is drawn in the plane of two suitable neural outputs.
    Preview · Article · Feb 2007 · Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference
  • [Show abstract] [Hide abstract]
    ABSTRACT: We set up a classification system able to detect patients affected by migraine without aura, through the analysis of their spontaneous EEG patterns. First, the signals are characterized by means of wavelet-based features, than a supervised neural network is used to classify the multichannel data. For the feature extraction, scale-dependent and scale-independent methods are considered with a variety of wavelet functions. Both the approaches provide very high and almost comparable classification performances. A complete separation of the two groups is obtained when the data are plotted in the plane spanned by two suitable neural outputs.
    No preview · Article · Aug 2007 · Physica A: Statistical Mechanics and its Applications
Show more