Realtime fMRI pattern-classification using artificial neural networks

We present a new approach to real-time analysis of fMRI data which allows to infer a subject's mental, perceptive or motor state from corresponding activation patterns during an ongoing fMRI session. This was achieved using a multivariate statistical procedure from the group of artificial neural networks (ANNs). As opposed to traditional voxel-wise statistics, multivariate procedures evaluate information contained in activation patterns distributed across the brain. These patterns correspond to the activity of multiple voxels. In the present study patterns of fMRI activation derived from single fMRI volumes were…