Epileptic Seizure Detection Based On Improved Wavelet Neural Networks

Epilepsy is a brain dysfunction disease and is chronic. So it is difficult to cure, and the patients often suffer from this disease for several years, even several decades. The characteristics of recurrent onset and uncertain time of onset bring serious damage to the patients’body and spirit. The families of the patients and society also bear huge burden and pressure. However, EEG, pattern recognition and signal processing technology have been developed well recently. As a result, the research on the seizure detection become more and more various. The researchers have obtained many important results recently. And automatic seizure detection could be of great importance for speeding up the inspection process and relieving the workload of medical staff in the analysis of EEG recordings.In this study, we have designed a new automatic seizure detection system. We have made an improvement on the wavelet neural network (WNN) in this system, and a method based on an improved WNN is proposed for automatic seizure detection in long-term intracranial EEG. First, WNN combines the traditional back propagation neural network (BPNN) with wavelet transform, and obtain advantage of assured convergence, ect. Then, compared with the classic WNN, a modified point symmetry-based fuzzy c-means (MSFCM) algorithm is applied to the initialization of wavelet transform’s translations, which has been successful in multiclass cancer classification. Fast-decaying Morlet wavelet is chosen as the activation function to make the WNN learn faster. In addition, relative amplitude and relative fluctuation index, which could denotes the EEG signals well, are extracted as feature vector to describe the variation of EEG signals, and the feature vector is then fed into WNN for classification. At last, post-processing including smoothing, channel fusion and collar technique is adopted to achieve more accurate and stable results.We have done an experiment with the designed system, and the system, which has taken advantage of every part well, performs efficiently with the average sensitivity of 96.72%, specificity of 98.91% and false-detection rate of 0.27/h. The proposed approach achieves high sensitivity and low false detection rate, which demonstrates its potential for clinical usage.