This project aims to evaluate nonlinear methods that enable analyzing segments of interest in a record EEG surface of long duration (typically 6 hours), taken in patients with refractory epilepsy, associated with the presence of High Frequency Oscillations (HFO: 40 - 100 Hz). To minimize noise and artifacts, are considered signs taken during the patients' sleep. The detection of OAFs be performed in a semiautomatic through the time-frequency transform, or need the detection of visual evaluation end of an expert, however aiming to minimize the maximum latter task. Then nonlinear quantifiers will be evaluated in segments with and without HFO, and comparatively EEG segments of normal individuals. The signals to be evaluated are from the database maintained by neurologists linked to Surgery Center in Epilepsy (CIREP), Hospital de Clinicas - USP, Ribeirão Preto. The nonlinear analysis or quantifications used in this study are: qSDiff Entropy; Detrended Fluctuations Analysis (DFA); Multiscaled Entropy (MSE); Symbolic Dynamics (SD). It is expected in this study: to find out whether it is possible to differentiate HFO corresponding to the epileptogenic zone (EZ) from segments of the EEG signal channels sets that do not contain these oscillations, and evaluating nonlinear characteristics and properties associated with HFO associated to pathologic ones. This study will evaluate the nonlinear characteristics of HFO have clinical utility as biomarkers of epileptic seizures, as well as biomarkers of ZE. HFOs are objects of great attention of neurosurgeons, which need to find ZE to perform invasive procedures needed to free patients from uncontrollable crises of this type.
News published in Agência FAPESP Newsletter about the scholarship: