Research On Motion Artifact Rejection And Stabilization For ECG Monitoring And Brain Computer Interface

Currently,ECG and EEG systems are two of the major diagnosis tools for human bioelectric signals detection,which have been widely applied including in military,aerospace and medical research area.To meet different requirements of these bioelectric systems,wearable device is becoming more popular.Not like those tranditional devices,protable bio-medical systems have low-cost,easy-operation,intelligent and can constitute a feedback loop with people,which brings a ground challenge to establish a stable system under all use cases.For a high integrated and efficient design of the wearable device,non-invasive biological signals acquisition is a must requirement which usually can be done with body-surface-attached sensing.However,when the observer is in motion such as eye blink or a deep breath,motion artifact(MA)can cause tremendous errors due to the time-variant skin contact,resulting in a low signal-to-noise ratio,large signal distortion,and low spatial resolution.In order to correct and compensate the motion artifacts,there are two aspects of research interests.On one hand,previous studies have focused on the electrode-skin contact layer(ETI)modeling and independent variable analysis.These research are based on a certain type of electrode for ETI modeling,we can only estimate the artifact based on one specific electrode parameters.Although this is conducive to the observation of the ETI equivalent impedance changes,but the correctness of the distortion signal depends entirely on the ETI model accuracy and electrode type,which can’t guarantee the versatility.On the other hand,many researchers are using independent component analysis algorithms to separate or extract motion artifacts.This method is badly time-consuming and requires a lot of repetitive experimental data,which is a burden for both subjects and researchers.To solve these technical difficulties discussed above,this work presented a novel motion artifact rejection and stabilization(MARS)technique for biomedical signal acquisition systems.Compared to traditional research,the author started from general model of ETI,to release the strict requirements of RC parameters of ETI network,it has to establish a reliable relationship between system transit equation and signal distortion.Based on the equation,the author first present the concept of multiplicative motion artifact,which supplement the knowledge of motion artifacts.Previously,people only study the baseline wandering issue caused by motion artifact.This work represents the motion artifacts as DC and AC factors,they are AMA and MMA.Aiming at these two different types of motion artifacts,a digital background adaptive compensation algorithm based on pseudo-random sequence correction technique is designed.The author involved the digital background calibration technique-PN sequence theory into bio-medical application.The proposed MARS system can achieve simultaneous correction for these two artifacts,so as to achieve high-accurate signal recovery.Since the electrode placement and circuit structure are different for ECG and EEG acquisition systems,this work established two test benches.One is integrated circuit design for MMA tracking module,this ASIC was designed for ECG recording,which is using 65 nm CMOS technology.Another test bench was designed for both ECG and EEG systems,it is based on Xilinx Zynq-7000 EVM board and programmable TI ADS1299 bio-medical signal acquisition board.As the electrode material is different,the equivalent impedance of the electrodeskin contact layer will also have corresponding variation.Meanwhile,different types of electrodes will have different baseline drift due to piezoelectric effects.Hence,those uncertain factors can easily degrade the dynamic response of the acquisition circuit,which increase the signal distortion.The theory verification and hardware implementation were qualified in University of Texas at Dallas and University of California at San Diego.In order to verify the versatility of the proposed techniques,two electrodes were selected for ECG and EEG observations during hardware implementation and test phase,including foam-type wet electrodes for ECG testing and metal dry electrodes for EEG detection.During real-time experiments,the distorted ECG signal,alpha brain waves,and SSVEP brain waves are used for the tests,which also contain the motion artifacts.Based on the MARS technique,the motion artifact under these three different experimental conditions is compared,which proves that the multiplicative artifacts are real and have a high degree of time consistency with the action.After correction,the ECG signal baseline drift is completely removed and QRS complex wave expression is accurate,while the disappearance of the T wave is restored.The calibrated EEG data could distinguish alpha brain waves between open eyes and closed eyes,and the target signal frequency range is correct.SSVEP brain waves are induced by 10 Hz flicker,and 20 Hz and 30 Hz are completely removed after correction.Thus,the MARS technology breaks the current experiment limitation,allows a higher tolerance to motion artifact and significantly improves the effectiveness and versatility of ECG/EEG acquisition.In sum,the proposed MARS technology release the strict requirements for ECG/EEG recording in laboratory environment.The subject will not have to remain stationary during test.It is compatible with arbitrary electrodes or systems in the market,which does not need any pre-acknowledge about the parameters of the electrode.The benefit of this work is avoiding complex analog front-end circuit design and back-end signal processing algorithms.Moreover,this work can track the MA in a single test to avoid repetitive measurements.Lastly,MARS also helps to monitor biomedical signals on moving subjects with higher accuracy.These results qualified the effectiveness and robustness for motion artifacts rejection.