(Pontificia Universidad Católica del Perú, 2018-03-20) Petersamer, Matthias; Haueisen, J.
The aim of this Master's Thesis was to develop a naturally controllable BCI that can predict
motion trajectories from the imagination of motor execution. The approach to reach this aim
was to nd a correlation between movement and brain data, which can subsequently be used
for the prediction of movement trajectories only by brain signals. To nd this correlation, an
experiment was carried out, in which a participant had to do triggered movements with its right
arm to four di erent targets. During the execution of the movements, the kinematic and EEG
data of the participant were recorded. After a preprocessing stage, the velocity of the kinematic
data in x and y directions, and the band power of the EEG data in di erent frequency ranges
were calculated and used as features for the calculation of the correlation by a multiple linear
regression. When applying the resulting regression parameter to predict trajectories from EEG
signals, the best accuracies were shown in the mu and low beta frequency range, as expected.
However, the accuracies were not as high as necessary for control of an application.
