Abstract #3685

Using Real-Time FMRI to Control a Dynamical System

Anders Eklund^1,2, Henrik Ohlsson³, Mats Andersson^1,2, Joakim Rydell¹, Anders Ynnerman^2,4, Hans Knutsson^1,2

¹Div. of Medical Informatics, Dept. of Biomedical Engineering, Linkping University, Sweden; ²Center for Medical Image Science and Visualization (CMIV), Linkping, Sweden; ³Div. of Automatic Control, Dept. of Electrical Engineering, Linkping University, Sweden; ⁴Div. for Visual Information Technology and Applications, Dept. of Science and Technology, Linkping University, Sweden

We present an fMRI based brain computer interface (BCI). The brain and the computer are linked by fMRI and work as a controller for a dynamical system. The dynamical system consists of a cart with an inverse pendulum mounted on it. A neural network is used to classify between rest, left hand and right hand activity. The classification computed by the neural network is used as an input to the dynamical system. The BCI runs in real-time and a new control signal is computed once a second. The subject was able to balance the inverse pendulum for 7 minutes.