International Journal of Bio-Science and Bio-Technology Vol.8, No.5 (2016), pp. 139-154 http://dx.doi.org/10.14257/ijbsbt.2016.8.5.13 ISSN: 2233-7849 IJBSBT Copyright ⓒ 2016 SERSC Effective EEG Motion Artifact Removal with KS test Blind Source Separation and Wavelet Transform Vandana Roy and Shailja Shukla DoEC, GGITS, Jabalpur, M.P., 482005, INDIA Professor and Head of DoCSE, JEC, Jabalpur, MP, 482011, INDIA Email: vandana.roy20@gmail.com, shailja270@gmail.com Abstract Artifacts frequently corrupt biomedical signal recording and processing, therefore, removal of these artifacts from physiological signals is an essential step. The acuteness in the performance of healthcare technology has upgraded from the current hospital-centric environment towards portable ubiquitous approaches. The uncertainty in the subsequent performance of these approaches introduced a dedicated research and past few decades have witnessed considerable improvement. In this research work an enhanced empirical approach to model the artifacts of EEG signal are described. The input EEG is a single channel and is converted into multichannel using Ensemble Empirical Mode decomposition (EEMD) operations and further filtered with Independent Component Analysis and Double Density Wavelet Transform to reject any traces of artifacts left at signal. This proposed algorithm is tested with different evaluation parameters and results pronounce the eligibility of the proposed algorithm to stand on top of currently deployed algorithms because significant improvement in results. Keywords: EEG, EMG, EEMD-ICA, ICA, DDWT, EEMD-DDWICA 1. Introduction The design of a physiological signal is a long followed approach that illustrates the present condition of the health of an individual. The dynamic research in medical sciences towards superlative health assessment calls for paramount accuracy with low computing cost in signal recordings and imaging. The life of an instrument to function with minimal operation and maintenance cost is one of the primary factors that defines the probability of its selection as the practically acceptable technology. Also, the simplicity of an instrument is directly measured as its capability to function in a standard environment without failure in operation. Further, the complexity of instruments has direct relationship with its cost. It is evident that measurement of physiological signals in even the surgical environment is accustomed to some noise also referred as artifacts in medical terms. These artifacts are unwanted signals generated due to unregulated sources besides the source under consideration. The artifacts in neural signals have two prominent sources other than the machine and environment noise. The muscular and ocular activities of an individual generate electric pulses of low amplitude and frequency that falls in filter range of sensors and recording equipment. Hence, artifacts rejection is a fundamental subject of research and is well researched [[1][1]]. This paper considers the artifacts caused due to the motion. Till this point, numerous applications of Independent Component Analysis (ICA), wavelets, and adaptive filters are proposed in the same context of research [[2]]. Normally, a common approach is to reject all Electroencephalogram (EEG) epochs containing the signal amplitude larger than some selected value. These methods are inflexible and do not allow for any adaption, which causes in loss of a portion of meaning full data. A component based automated separator of artifacts is required to overcome this