- NB: Document concerns OpenViBE >= 0.18.0
This set of scenarios implements the Graz BCI, based on motor imagery of the hands. It computes the spatial filters that efficiently discriminate the signal using CSP, for significantly better performances. You will find it in share/openvibe/scenarios/bci-examples/motor-imagery-CSP.
- mi-csp-0-signal-monitoring.xml: This scenario should be always used prior to anything and in background to check the signal quality of the acquisition device. Once you are sure that the EEG acquisition runs correctly, you can go on to the next step !
- mi-csp-1-acquisition.xml: First step is to acquire some data in order to train the classifier that will discriminate Right and Left hand movements. The training session can be configured in the LUA stimulator (number of trials, timings, etc.).
- mi-csp-2-train-csp.xml: This scenario computes Common Sptial Pattern to produce a spatial filter that maximizes the difference between the signal of the two classes. Use a previously acquired file to perform the training.
- mi-csp-3-classifier-trainer.xml: This scenario trains a LDA classifier based on the previous acquisition session. Note that the signal processing pipeline may be tuned accroding to the type of data acquired. For example, the Reference Channel may not be needed.
- mi-csp-4-online.xml: This scenario adds real-time feedback to the visualization, using the trained LDA classifier. Again, you may have to tune the signal processing pipeline.
- mi-csp-5-replay.xml: This scenario is based on the online one, but the input signal is coming from a file rather than acquisition server. In this version classifier performance tools are used to display the confusion matrix of the classifier and its global performance during the session. Note the existence of the stream switch box. This does not currently exist in the corresponding online scenario. It is used to get accurate measurements of during-trial data only.