Research And Implementation Of Brain Wave Anomaly Detection Based On Deep Learning

Epilepsy is a disease caused by abnormal brain discharges that not only causes great suffering but also poses a great threat to the daily health of patients.How to predict future epilepsy diseases has become an urgent need to solve important question.In recent years,with the rapid development of computer technology and hardware facilities,deep learning technology has made more and more achievements in the medical field.Some of the problems that plague doctors,such as the diagnosis of arrhythmia types and the diagnosis and determination of lesion locations,can be well accomplished by deep learning technology.In this thesis,we try to study how to process the epilepsy data in the epilepsy EEG database in the most efficient way so that the data can be used by the neural network most effectively,and try to process the data by drawing the EEG signal into EEG based on the traditional methods of processing epilepsy EEG data,and use the sliding window method to further improve the information and effectiveness of the data based on these EEG data.Secondly,to address the three characteristics of picture-based,temporal and biological nature of epileptic EEG training data,this thesis uses epileptic EEG data to train CNN,LSTM and SNN network structures,and makes further improvements and studies according to the characteristics of these three types of network structures.Thirdly,in response to the lack of data and data imbalance in the current deep learning technology using epilepsy EEG data for training,this thesis first analyzes the commonly used base data augmentation methods and tries to conduct data augmentation experiments using GAN,which has excellent performance in the field of image generation.Finally,this thesis summarizes the work in the above research and experiments as well as the shortcomings and deficiencies in them,and elaborates the directions for improvement based on these shortcomings and deficiencies,and also makes an outlook on the further application of deep learning techniques in the field of epilepsy diseases after finishing all the work.