Keshri Anup Kumar, Sinha Rakesh Kumar, Hatwal Rajesh, Das Barda Nand
Department of Computer Science & Engineering, Birla Institute of Technology, Mesra, Ranchi, Jharkhand 835215, India.
J Med Syst. 2009 Jun;33(3):173-9. doi: 10.1007/s10916-008-9177-1.
This Paper presents an automated method of Epileptic Spike detection in Electroencephalogram (EEG) using Deterministic Finite Automata (DFA). It takes prerecorded single channel EEG data file as input and finds the occurrences of Epileptic Spikes data in it. The EEG signal was recorded at 256 Hz in two minutes separate data files using the Visual Lab-M software (ADLink Technology Inc., Taiwan). It was preprocessed for removal of baseline shift and band pass filtered using an infinite impulse response (IIR) Butterworth filter. A system, whose functionality was modeled with DFA, was designed. The system was tested with 10 EEG signal data files. The recognition rate of Epileptic Spike as on average was 95.68%. This system does not require any human intrusion. Also it does not need any short of training. The result shows that the application of DFA can be useful in detection of different characteristics present in EEG signals. This approach could be extended to a continuous data processing system.
本文提出了一种使用确定性有限自动机（DFA）在脑电图（EEG）中自动检测癫痫棘波的方法。它将预先记录的单通道EEG数据文件作为输入，并在其中找到癫痫棘波数据的出现情况。使用Visual Lab-M软件（台湾研华科技股份有限公司）在两个单独的两分钟数据文件中以256Hz记录EEG信号。对其进行预处理以消除基线漂移，并使用无限脉冲响应（IIR）巴特沃斯滤波器进行带通滤波。设计了一个用DFA对其功能进行建模的系统。该系统用10个EEG信号数据文件进行了测试。癫痫棘波的平均识别率为95.68%。该系统不需要任何人工干预。此外，它也不需要任何形式的训练。结果表明，DFA的应用在检测EEG信号中存在的不同特征方面可能是有用的。这种方法可以扩展到连续数据处理系统。
