“School of Cognitive”

Back to Papers Home
Back to Papers of School of Cognitive

Paper   IPM / Cognitive / 12327
School of Cognitive Sciences
  Title:   Predicting the cognitive states of the subjects in functional magnetic resonance imaging signals using the combination of feature selection strategies
  Author(s):  Mohammad Reza Daliri
  Status:   Published
  Journal: Brain Topogr
  Vol.:  25
  Year:  2012
  Pages:   129-135
  Supported by:  IPM
  Abstract:
The functional magnetic resonance imaging (fMRI) provides very useful information about the activities from different brain areas during a task. This information can be used to train a classifier and predict the sensory and motor functions and also different mental states of the subject?s brain in a particular task. Using a high resolution fMRI, normally the activities from many voxels are obtained with respect to time and not all of these voxels involve actively in a particular task. Here we propose a combination of feature selection strategies using an evolutionary computation algorithm and the support vector machines to find out those feature dimensions that are actively involved in representing the brain activities in a particular task. We show that using this lower dimensional space we can predict the cognitive state of the subjects in a particular task more accurately.

Download TeX format
back to top
scroll left or right