Research On Automatic Seizure Detection Method Based On PLV Brain Network

Epilepsy is a common neurological disorder characterized by transient loss of consciousness,muscle reflexes,and other symptoms,which can cause great pain to patients.In clinical diagnosis and treatment,physicians usually rely on electroencephalogram(EEG)signals to assess the patient’s condition.However,this is a time-consuming and error-prone task.Therefore,the academic community has proposed various automatic seizure detection algorithms based on EEG time and frequency domain features.Traditional time and frequency domain features lack understanding and interpretation of internal information exchange in the brain.To address this issue,researchers have begun exploring the use of brain networks to obtain more interpretable features.Brain networks reflect the information exchange between different regions of the brain,which is more in line with the mechanism of a large amount of information exchange between neurons during brain function.Therefore,brain networks constructed using EEG can be used as one of the discriminatory bases for seizures,and also provide more interpretable features.Moreover,previous studies have found that the amount of information obtained by different brain network construction methods varies.Based on the above issues,this article proposes a brain network-based automatic seizure detection algorithm to meet clinical needs.The experiment was conducted on the CHB-MIT dataset of Boston Children’s Hospital,and the results show that the performance of the seizure detection algorithm is significantly improved after incorporating brain network features.The purpose of this article is to explore the brain network-based automatic seizure detection method,and the main research contents are as follows:(1)Aiming at the problem that most of the current brain network studies are based on EEG signal amplitude and ignore the importance of EEG signal phase,this paper proposes a method to combine EEG signal amplitude and phase information to build a brain network.This method constructs different brain networks based on the phase and amplitude of EEG signals,and then integrating these two networks to create a brain network that contains both amplitude and phase information.(2)Aiming at the lack of interpretability of traditional methods based on time and frequency domain features for seizure detection,an automatic detection algorithm based on brain networks and machine learning is proposed in this paper.The algorithm extracts features such as global efficiency and characteristic path length of the brain network,and combines them with traditional time-domain and frequency-domain features to construct a feature vector.Finally,the feature vector is input into a support vector machine classifier for epileptic seizure detection.The experimental results show that the proposed method improves the performance of epileptic seizure detection.(3)Aiming at the problem that time series features are missing in the seizure detection methods based on brain networks and machine learning,this paper proposes a deep learning seizure detection algorithm based on time series features of brain networks.This paper uses a sliding window for multiple feature extraction of data fragments,and then the extracted features are formed into time series features according to the direction of the sliding window.Finally,the time series features are input into classifiers such as convolutional neural networks and recurrent neural networks for epilepsy detection.The experimental results show that the sensitivity index of seizure detection can be significantly improved when adding brain network features.