Nonlinear analysis of EEG for epileptic seizures
- Oak Ridge National Lab., TN (United States)
- Knoxville Neurology Clinic, St. Mary`s Medical Center, Knoxville, TN (United States)
We apply chaotic time series analysis (CTSA) to human electroencephalogram (EEG) data. Three epoches were examined: epileptic seizure, non-seizure, and transition from non-seizure to seizure. The CTSA tools were applied to four forms of these data: raw EEG data (e-data), artifact data (f-data) via application of a quadratic zero-phase filter of the raw data, artifact-filtered data (g- data) and that was the residual after subtracting f-data from e-data, and a low-pass-filtered version (h-data) of g-data. Two different seizures were analyzed for the same patient. Several nonlinear measures uniquely indicate an epileptic seizure in both cases, including an abrupt decrease in the time per wave cycle in f-data, an abrupt increase in the Kolmogorov entropy and in the correlation dimension for e-h data, and an abrupt increase in the correlation dimension for e-h data. The transition from normal to seizure state also is characterized by distinctly different trends in the nonlinear measures for each seizure and may be potential seizure predictors for this patient. Surrogate analysis of e-data shows that statistically significant nonlinear structure is present during the non-seizure, transition , and seizure epoches.
- Research Organization:
- Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- AC05-96OR22464
- OSTI ID:
- 366563
- Report Number(s):
- ORNL/TM-12961; ON: DE96013818
- Resource Relation:
- Other Information: PBD: Apr 1995
- Country of Publication:
- United States
- Language:
- English
Similar Records
Apparatus and method for epileptic seizure detection using non-linear techniques
Apparatus and method for epileptic seizure detection using non-linear techniques