Page 1 of 1

Cursor Task problems

Posted: 03 Jul 2010, 09:47
by engicri
Hi to all!

I have some problems with the BCI2000 Cursor Movement Task application and I hope someone could help me.

The same user performed the Stimulus Presentation task four times in three days. The protocol described in the Mu Rhythm tutorial was implemented, considering the 4 motor tasks left hand , right hand , both hands and both feet movements. Considering the corresponding r^2 value in the feature plot and the topography plot in the offline analysis, the best task was the LEFT HAND, .

Then the user performed the cursor task application, to test the only left hand movement to horizontally move the cursor toward the left screen side.
The CAR filter was adopted as spatial filter and the classifier matrix was defined following the protocol described in “Using an EEG-Based Brain-Computer Interface for Virtual Cursor Movement with BCI2000” (J. Adam Wilson, Gerwin Schalk, Léo M. Walton, Justin C. Williams) (weight -1 for the channel C3, +1 for the channel C4).
Nevertheless no good results have been obtained. Without any task performing, the cursor moves towards right. When the user moves his left hand, the cursor doesn't follow a clear direction and as the user stops to move his hand, the cursor moves spontaneously toward the right edge.
I increased the weight in the classifier matrix up to 10 and a little better control of the cursor movement by the left hand has been obtained but it isn't enough good.

Where can the problem be? Every suggestion will be really appreciated.

Thanks in advance,

Posted: 08 Jul 2010, 08:19
by mellinger
Make sure that the Filtering->Adaptation parameter is set to 2 for the horizontal direction (first entry), and that the Filtering->BufferConditions parameter contains "Feedback&&(TargetCode==1)" and "Feedback&&(TargetCode==2)" as conditions in its first column.

Also, perform a number of trials to give the adaptation a chance to work.