Research On Hand Rehabilitation System Of Stroke Based On Deep Learning

Machine learning algorithms have appeared as early as the 1960 s,and with the rapid development of computer hardware,high-performance computing has been realized.Therefore,in recent years,deep learning algorithms,as an important branch of machine learning,have been rapidly developed,with outstanding performance in all fields.In health care,deep learning is also used for a variety of clinical practical problems.Hemiplegia sequelae caused by stroke has always been a difficult problem for patients and medical staff.Therefore,this paper put forward a based on the deep learning method to identify stroke patients EEG signals,actively induce patient rehabilitation training,and deep analysis in deep learning model identification based on the principle of patients' hand movements to imagine,puts forward the construction of the recovery system of a complete set of processes,so as to achieve the auxiliary medical personnel to the effect of rehabilitation treatment to patients.This paper mainly carries out relevant research from the following three aspects,and the main work is summarized as follows:Firstly,based on the background of this study,the harm and existing rehabilitation methods of hemiplegic motor dysfunction caused by stroke disease were introduced,and the existing problems were pointed out.By introducing the rehabilitation therapy based on BCI sports imagination in detail,the basic principle and research value of the research are expounded.Secondly,the mechanism of EEG signal generation and the information contained in it are used to introduce the generation of motion imagination process and the phenomenon of event correlation desynchronization/synchronization(ERD/ERS)in detail.The STFT and HHT algorithms which are the most widely used and most effective in EEG signal processing are proved theoretically.Based on the signal processing theory,the signal enhancement technique used in this paper is discussed.Thirdly,from the stroke hand rehabilitation training system to build the process.Through the working process of the system,the whole structure and construction method of the rehabilitation system are briefly introduced.Starting from the system software and hardware,this paper mainly introduces the design ideas and construction methods of EEG signal acquisition,signal processing,algorithm model,control system and control equipment.Through the problems encountered in the actual test,the rehabilitation system was optimized and improved,and the design concept and the expected goal of the rehabilitation system of the hand for stroke were initially realized.Finally,in the algorithm module used to identify the EEG signals of patients' motion imagination,the design process of the experimental paradigm of motion imagination and the three data sets used in model training are introduced in detail.When the model algorithm is selected,several motion image recognition algorithms are used to test the data set,and the classification effect is compared.The performance of different models in data enhancement technology is compared with the classification effect of original data set to verify the reliability of the technology.