Basic Study to Reduce the Artifact in Brain Activity Measuring Data by Using Auto-Regressive Model

Abstract

Near Infra-Red Spectroscopy (NIRS) is possible to measure brain activity signals with low invasive and low physical restraint, so it is also expected as a useful clinical tool in the fields of medicine and education. However, measurement and detection of weak signal of brain activities such as human intention and recall of memory, is very difficult because noise signals such as heartbeat, respiration, and body movement disturb the brain activity signal and data. In this study, we attempt to reduce the effect of the artifacts ingredient by the body movement from the data of NIRS using the Auto-Regressive (AR) model. From the experimental results, we confirmed that the peaks of FFT analysis of brain activity measurement data and ECG data were almost identical. Particularly, characteristic peaks around 0.5Hz and 1.3Hz were detected. The peak around 0.5Hz was effected based on position change of the head inclining. The peak around 1.3Hz was effected based on the periodical R wave in the ECG data. Also, AR model was estimated from AR coefficients by the above results. So, we could separate the component of artifacts based on heartbeats and separate the component of artifacts based on positioning change of the head inclining, from the brain activity signals. It means that using AR model from the brain activity measurement data by NIRS, we could reduce the effects of artifacts based on heartbeats and reduce the effects artifacts based on position change of the head inclining. In the future, we will reduce other artifacts based on other factors. Moreover, using this proposed methodology, we try to improve brain region identification and statistical accuracy which could not be clarified by conventional experimental methods.