A Comparison of Actual and Artifactual Features Based on Fractal Analyses: Resting-State MEG Data

Montri Phothisonothai,Hiroyuki Tsubomi,Katsumi Watanabe,Yoshio Minabe,Yuko Yoshimura,Mitsuru Kikuchi,Aki Kondo

doi:10.1007/978-3-642-37502-6_146

Montri Phothisonothai, Hiroyuki Tsubomi + Show 5 more

https://doi.org/10.1007/978-3-642-37502-6_146

Copy DOI

Abstract

Future standardized system for distinguishing actual and artifactual magnetoencephalogram (MEG) data is an essential tool. In this paper, we proposed the quantitative parameters based on fractal dimension (FD) analyses in which the FD may convey different features before and after artifact removal. The six FD algorithms based on time-series computation, namely, box-counting method (BCM), variance fractal dimension (VFD), Higuchi’s method (HM), Kazt’s method (KM), detrended fluctuation analysis (DFA), and modified zero-crossing rate (MZCR) were compared. These approaches measure nonlinear-behavioral responses in the resting-state MEG data. Experimental results showed that the FD value of actual MEG was increased statistically in comparison with the artifactual MEG. The DFA and the HM present a best performance for analyzing simulated data and resting-state MEG data, respectively.

Full Text