Abstract #1813
Automatic resting-state fMRI Independent Component Classification using Support Vector Machines
Yanlu Wang 1,2 and Tie-Qiang Li 1,2
1
Clinical Science, Intervention, and
Technology, Karolinska Institute, Stockholm, Stockholm,
Sweden,
2
Medical
Physics, Karolinska University Hospital, Huddinge,
Stockholm, Sweden
To facilitate the identification of meaningful
components from ICA analysis for resting-state fMRI
data, we have developed a supervised classification
framework based on support vector machines for automatic
identification of noise/artifact components. By using
classifiers that reflect typical instructions for visual
inspection and are invariant of training dataset, our
framework achieved zero false negative rates and
consistently low false positive rates for identifying
noise/artifact components. Our framework facilitates
ICA-based analysis of resting-state fMRI data with high
model orders, and can be used for automatic removal of
noise/artifact components without risking discarding any
potentially interesting and meaningful components.
This abstract and the presentation materials are available to members only;
a login is required.
Join Here
