Eeg Analysis By Using High-performance Computation

The human brain consisting of about 140 trillion nerve cells is one of the most complicated organs in human body. EEG is the result of many neurons'joint activities, which reflects electrophysiological movements of brain nerves. Research of EEG, once found, integrates with its applications tightly. At present, EEG studies have not only used in the diagnosis of disease but also had a positive effect in some functional rehabilitation. Therefore, the signal processing analysis methods of EEG is particularly important. As a very large amount of EEG data, processing is difficult, time consuming, so consider how effectively and quickly analyze and process the EEG signals and get relevant results become the focus of this article.This paper use a scheme combining algorithm with program development. First,for a group of anesthesia EEG data are processed and analyze by EEMD algorithms which is a novel adaptive time-frequency analysis method that is particularly suited to extracting useful information of noisy nonlinear or non-stationary processes,such as electroencephalogram (EEG) data. However EEMD algorithm is highly compute-intensive,so we apply high-performance computing to accelerate it. The accelerated computing algorithm is completed after that GPU-based Compute Unified Device Architecture which launched by the NVIDIA company is used in this paper. The main work is to analyze and process the EEG signals which useful information is obationed, then to extract its high density and complex computing task, and apply the high performance computing for the purpose of achieving rapid and accurate analysis results.Then, spatiotemporal dynamics is adopted to analyze epileptic EEG, the main purpose is discrimination for epileptic seizure's beginning, middle, end, applied to clinical evaluation and treatment. The results show that high-performance parallel computing has much high operation speed on the analysis of EEG signal processing, and greatly reduces the signal analysis and processing cycle, under the situation that the accuracy is not influenced.at the same timespatial temporal dynamic mechanism of seizures staging recognition has higher accuracy.