Projekts

Project II:
Software development

started 2007

The BMI Software is a C++ application for runtime decoding movement intentions from brain signals.
Different amplifiers can be used to record e. g. EEG or ECOG. If desired, the user’s movement with a joystick, mouse or haptic device can also be captured.

A typical use case has two phases.

In the first phase, brain signals for known movements are recorded. From this data a model for online decoding is derived.
During second phase this model is used to decode brain signals at runtime. The user can see the effect of its decoded movement in a paradigm view.

Currently we are focusing on:

• the incorporation of new machine learning techniques in order to make our decoding models more adaptive
• performance optimization

Sample references:

• Blumberg J, Rickert J, Waldert S, Schulze-Bonhage A, Aertsen A, Mehring C (2007) Adaptive Classification for Brain Computer Interfaces. Conf Proc IEEE Eng Med Biol Soc. 2007;1:2536-9.
• Rickert J, Braun D, Mehring C (2007), Unsupervised adaptive kalman-filter for decoding non-stationary brain-signals. Neuroscience 2007, Presentation Number:414.13