Dynamic feature extraction of epileptic EEG using recurrence quantification analysis

Detecting the reliable transition point embedded in the electroencephalograms (EEGs) is a challenge in the field of epileptic research. In this research, a recurrence quantification analysis (RQA) is proposed to help medical doctors to reveal dynamical characteristics in EEGs of patients suffering from epilepsy. In contrast with traditional chaos methods, the merits of RQA method is that it can measure the complexity of a short and non-stationary signal without any assumptions such as linear, stationary and noiseless noise. In this study, EEGs with generalized epilepsy were collected in Epilepsy Center of Renji Hospital. The test results show that three RQA measurements, i.e. recurrence rate, determinism and entropy can track the complexity changes of brain electrical activity. RQA variables show a large fluctuation in pre-ictal stage, which reflects a transitional state leading to seizure activity. On the contrary, RQA variables fluctuate in relatively small bounds in ictal stage, which is due to organized and self-sustained rhythmic discharge. Therefore, RQA could be a promising approach in prediction and diagnosis for epileptic seizures.