Automatic seizure detection based on time-frequency analysis and artificial neural networks

Abstract

The recording of seizures is of primary interest in the evaluation of epileptic patients. Seizure is the phenomenon of rhythmicity discharge from either a local area or the whole brain and the individual behavior usually lasts from seconds to minutes. Since seizures, in general, occur infrequently and unpredictably, automatic detection of seizures during long-term electroencephalograph (EEG) recordings is highly recommended. As EEG signals are nonstationary, the conventional methods of frequency analysis are not successful for diagnostic purposes. This paper presents a method of analysis of EEG signals, which is based on time-frequency analysis. Initially, selected segments of the EEG signals are analyzed using time-frequency methods and several features are extracted for each segment, representing the energy distribution in the time-frequency plane. Then, those features are used as an input in an artificial neural network (ANN), which provides the final classification of the EEG segments concerning the existence of seizures or not. We used a publicly available dataset in order to evaluate our method and the evaluation results are very promising indicating overall accuracy from 97.72% to 100%.

Figure 1

The flowchart of the proposed method.

Figure 2

Exemplary EEG segments from each of the five subsets (Z, O, N, F, and S). From top to bottom: subset Z to subset S. The amplitudes of surface EEG recordings are typically in the order of some μV. For intracranial EEG recordings, the amplitudes range around 100 μV. For seizure activity, these voltages can exceed 1000 μV.

Figure 3

The obtained spectrum for five EEG segments, one for each of the original dataset categories (Z, O, N, F, and S).

Figure 4

The spectrums obtained for various combinations of time and frequency partitions: (a) 3 time windows and 4 frequency subbands, (b) 5 time windows and 4 frequency subbands, (c) 3 time windows and 5 frequency subbands, (d) 5 time windows and 5 frequency subbands, (e) 3 time windows and 7 frequency subbands, (f) 5 time windows and 7 frequency subbands, (g) 3 time windows and 13 frequency subbands, and (h) 5 time windows and 13 frequency subbands.