Epilepsy Prediction Through Multi-dimensional Feature Analysis Of EEG Signal

Epilepsy is one of the most common neurological diseases.There are about 60 million epileptic patients in the world.Epilepsy is a sudden disease,once the seizure is not timely and effective treatment,it is very likely to be life-threatening,so it is very necessary to predict seizures.Epileptic seizures are caused by abnormal discharges of neurons in the brain,so electroencephalography(EEG)is the most commonly used method for monitoring,diagnosing and managing epilepsy related neurological diseases.Because of its advantages in biological signal processing,entropy method has become a hot spot in the study of epileptic EEG signal.The main research work of this paper is as follows(1)Based on the sample entropy(SampEn),the concept of Multi-dimensional sample entropy(M-SampEn)is proposed as the feature of epileptic seizure prediction.EEG acquisition instrument consists of multiple channels,so it is a feasible method to combine the EEG signals of multiple channels to construct M-SampEn and use it as the characteristic value of epileptic seizure prediction.The experimental results show that compared with SampEn,M-SampEn has improved performance in all aspects,reaching an average classification accuracy of 85.96% and a recall rate of 90.30%.(2)The method of M-SampEn is optimized.Due to the large amount of data used in the calculation of M-SampEn,this paper optimizes the algorithm principle and dimension reduction.The main idea of algorithm principle optimization is to combine repeated calculation;the main idea of dimension reduction is to locate the focus by calculating the energy and select the channel with great changes before and after the onset.The optimized M-SampEn algorithm improves the computational efficiency.When the signal sampling point is 4096,the computational efficiency is 9.722 times of the original,which is more suitable for clinical application.(3)Bi directional long short term memory networks(BI LSTM)was introduced to predict epileptic seizures.In this paper,Bi LSTM is used to predict the M-SampEn feature in the next period of time,and judge whether the feature is seizure feature,so as to achieve the purpose of predicting seizures.The average prediction accuracy is80.09%.This method can not only predict the onset,but also the duration of seizure.