A Study On The Electrical Stimulation System Of Evoked EEG Signals And The Associated Signal Processing Algorithms

Evoked potential (EP) refers to the corresponding changes of brain potentials caused by stimulating light, sound and electricity to the sense organs of human beings. Analysis of evoked brain potential is an important means to study brain function and brain cognitive activities. Body feeling evoked constant-current electrical stimulator (BFECCES) plays a significant role in the research area of EP. On the one hand, the research of BFECCES has vital significance in the areas of brain and nerve science, clinical diagnosis and rehabilitation medicine. On the other hand, it can strengthen the capability of independent innovation for medical equipments of bio-electrical detection in China. A qualified stimulator applied to evoke EEG acquisition appliances must meet many requirements including low power comsumption, superb anti-powerline interference performance, high accuracy and abudance of functions. Moreover, the safety of stimulator is the most important factor and remains a challenge in the design of a constant-current electrical stimulator. Aiming at enhancing the security and accuracy of BFECCES, in this dissertation we will conduct studies on both the hardware design and signal processing algorithm development.1. The first part focuses on optimizing the hardware design of FPGA control circuit in evoke EEG acquisition system according to the design requirements of low power consumption, high precision of current control and other related indicators with an emphasis on addressing the security problem. To prevent the stimulator from burning the skin, which happens when the width of stimulation pulses is too wide or the strength of current too high, we firstly design a pulse width limiting circuit capable of limiting the pulse duration in a time sale of one millisecond. This circuit consists of an FPGA as the controller and a 12-bit high-precision analog-to-digital conversion chip that can control the logical sequence of stimulating patterns effectively. Floating ground technology is employed in order to guarantee the safety requirement. Specifically, a diode/capacitor based boost circuit is adopted after the isolation transformer. This design can prevent the stimulation circuit from injecting constant current into the human body too long, and hence protect the human body from electric shock caused by strong current. In addition, we also design the periphery circuits of power-off protection, leakage protection, over-current protection for further safety assurance. Simulation results show that the designed stimulator satisfy all the original design requirements.2. The second part of this study employs independent component analysis (ICA), an advanced digital signal processing technique, to improve the quality of extracted body feeling evoked potential. For purpose of comparison, the ordinary extracting methods of evoked potential are also implemented. Experiment result show that feature extracted by ICA algorithm can overcome the adaptability error of nervous system caused by fatigue of the patient due to the time consuming experiment needed by ordinary methods. Compared with ensemble averaging method, the ICA method have moderate algorithmic complexity, fast processing speed and better performance in single extraction of different evoked potentials. Besides the algorithm design, we also develop an interface for the software of the body sense evoked potential signal analysis.The contribution of this dissertation is twofold. Firstly, we design a novel FPGA-driven body feeling constant-current stimulator enjoying many advantages such as low power consumption, high precision of current control and superb standard of safety. Secondly, we develop an ICA based algorithm for feature extraction of evoked potentials, which outperforms the ordinary ensemble averaging method.