The misdiagnosis rate of epilepsy is said to keep high from 5% to 30% because of dependency upon an individual judgment by each medical doctor in diagnosis and a quantitative index seems necessary to manage diagnosis uncertainty. To detect some change appearing in brain waves, we focus on introducing a Duffing oscillator model, which could provide reasonably good predictions for the dynamics of neuronal groups. The aim of this paper is to discuss indexes for epilepsy diagnosis by representing characteristics of electroencephalogram (EEG) quantitatively using a Duffing oscillator model. The model parameters are directly identified to adapt the characteristics of the temporal EEG variation to dynamical properties of the model quantitatively. Therefore, in animal experiments, we obtained time histories of the EEG data changed from normal EEG to the epileptic EEG. As a result, it is found that the parameter values related to non-linearity are extremely reduced as the epileptic EEG progresses with time. On the other hand, the input signal strength in epileptic EEG is much bigger than that of normal as expected. Moreover, the directly identified exciting frequency and the eigenfrequency determined by the identified parameter exist in wider band than that of normal as the epileptic EEG progresses with time. The change of the EEG due to epileptic seizure could reflect on the model parameters and it is shown that the model parameters have the possibility to use as supporting index of diagnosis about epilepsy. As a result, the proposed method could be used to support the decline of misdiagnosis rate of epilepsy.

This content is only available via PDF.
You do not currently have access to this content.