Software Development For Analyzing High-Throughput Neural Signal In Acquisition And Recording System

Aiming to study,locate,assist,enhance and repair human cognitive or sensory motor function,Brain Computer Interface(BCI)technology provides a new treatment method for patients who have lost motor function.It is of great significance to the research of brain science,which has applied widely in medical,entertainment,aerospace,military and other fields.With further development of brain science,the requirement of high-throughput neural signal acquisition and analysis in BCI system is also higher.Hence,the high-throughput neural signal acquisition and recording system is of high value of both scientific research and engineering application.In order to meet the needs of experimental research on sensory-motor neural signal analysis in non-human primates,this paper develops a high-throughput brain neural signal acquisition and recording system analysis software.By designing the architecture through modularization,the software realizes the functions of real-time acquisition,network storage,spike signal analysis,graphical display and so on.First,to work on neural signals' features of large data channels and huge amounts of data,the software introduces the method of serial number matching and data packet buffering,ensuring the synchronization and real-time of data collection.Second,by connecting to IPSAN through iSCSI protocol,it develops a data storage scheme that could separate index files and data files to meet the requirements of efficient storage,retrieval and payback of high-throughput neural signal data,which is based on NTFS file system.Third,in the aspect of spike signal analysis,By designing the algorithm module of signal classification and signal decoding(in a component-based way)and improving the performance of main algorithms(through CUDA),the software realizes the efficiency and flexibility of analysis and configuration Last,on the basis of the QT interface,it realizes the topological connection of algorithm components and the display of general waveform control which provides a friendly user Human computer interface.The test results show that the system could achieve all the expected functions of the software,including real-time acquisition,network storage,spike signal analysis,graphical display,etc.In terms of performance,according to the on-line test of simulated 1024 channel neural signal,the average time of spike classification in real-time operation of the system is less than 4ms,which meets the real-time requirements.According to the off-line test of animal experimental signals,it can complete the off-line analysis of the experimental data of monkey through 5 methods.Every method could reach average time of 6 milliseconds for single frame data decoding.The efficiency of analysis is increased by 40.8%with GPU parallel processing.It meets the requirements of animal experimental research.