Mohammad Asghari Jafarabadi - Dept. of Biostatistics, School of Medical Sciences, Tarbiat Modares University, Tehran, Iran
Ebrahim Hajizadeh - Dept. of Biostatistics, School of Medical Sciences, Tarbiat Modares University, Tehran, Iran
Anooshiravan Kazemnejad - Dept. of Biostatistics, School of Medical Sciences, Tarbiat Modares University, Tehran, Iran
Manoochehr Emadi - Pars Hospital, Tehran, Iran

Background: Providing a suitable representation of multivariate data based onlinear transformation is the goal of various fields of researches such as signalprocessing. Methods: For this reason, principal component analysis as a classicaltechnique and independent component Analysis as a new technique, used onMagneto encephalography records, however the optimality of these methods isn tknown for such data. The aim of this article is to compare the two methods andrepresent the optimal technique for Magneto encephalography signal processing inorder to separate the artifacts and noise from brain original signals. Results: Theapplication of these methods on the Magneto encephalography data show thatindependent component Analysis performs very better than principal component analysis and could well separate the muscle activity, saccade related, blinking,respiration and heart rate artifacts and noises from brain original signals.Conclusion: After identification and possible separation and removing of thesesignals, more accurate studies could be done on brain activities.

Principal Component Analysis, Independent Component Analysis, MEG, Artifact