Skip to main content
. Author manuscript; available in PMC: 2012 May 15.
Published in final edited form as: Neuroimage. 2010 Jun 25;56(2):517–524. doi: 10.1016/j.neuroimage.2010.05.065

Table 2.

The decrease in testing accuracy with artifact removal indicates that some of the classification ability is due to systematic noise in the scans.

Predictive Accuracy over Artifact Removed Data
Model Component Extraction Run Model Training Run Model Testing Run Overall Testing Accuracy Control Testing Accuracy Patient Testing Accuracy Artifact Removal Methods
G 1 1 1 41.67% 27.27% 53.85% FSL, Manual
H 2 2 2 79.17% 63.64% 92.31% FSL, Manual
I 2 2 2 75% 75% 75% FSL, Tohka,

“FSL”: artifact removal using standard routines in software package. “Tohka”: automated ICA classifier (within subject) and removal of artifactual components. “Manual”: remove group-wide ICA components associated with artifacts.