Data set Ib: ‹self-regulation of SCPs›
Datasets provided by University of Tuebingen.
Correspondence to Thilo Hinterberger
<thilo.Hinterberger@uni-tuebingen.de>
Description of Experiments:
The datasets were taken from an artificially respirated ALS patient. The
subject was asked to move a cursor up and down on a computer screen, while
his cortical potentials were taken. During the recording, the subject received
auditory and visual feedback of his slow cortical potentials (Cz-Mastoids).
Cortical positivity lead to a downward movement of the cursor on the screen.
Cortical negativity lead to an upward movement of the cursor. Each
trial lasted 8s.
During every trial, the task was visually and auditorily presented
by a highlighed goal at the top (for negativity) or bottom (for positivity)
of the screen from second 0.5 until second 7.5 of every trial. In
addition, the task ("up" or "down") was vocalised at second 0.5.
The visual feedback was presented from second 2 to second 6.5. Only
this 4.5 second interval of every trial is provided for training and testing.
The sampling rate of 256 Hz and the recording length of 4.5s results in
1152 samples per channel for every trial.
Trial structure overview:
Duration: 8 s, no inter trial intervals
task presentation: from 0.5 s to 7.5 s
feedback period: from 2.0 s to 6.5 s
Data:
Amplifier: PsyLab EEG8
A/D-converter: Computer Boards PCIM-DAS1602/16 bit
Amplitude range: +/-1000 µV
Sampling rate: 256 S/s
EEG data was taken from the following positions:
Channel 1: A1-Cz (10/20 system) (A1 = left mastoid)
Channel 2: A2-Cz
Channel 3: 2 cm frontal of C3
Channel 4: 2 cm parietal of C3
Channel 5: vEOG artifact channel to detect vertical eye movements
Channel 6: 2 cm frontal of C4
Channel 7: 2 cm parietal of C4
All values are in µVolt.
The EEG-values are not EOG-corrected.
Traindata_0.txt and Traindata_1.txt:
These files contain 200 trials which were recorded on the same day and
permuted randomly.
The 2-D Matrices Traindata_0.txt and Traindata_1.txt each contain data
of 100 trials belonging to class the corresponding class. The matrix dimensions
are 100x8065 and 100x8065.
Every line of a matrix contains the data of one trial. The first column
codes the class of the trial (0/1). The remaining columns contain the time
samples of the 7 EEG/EOG channels. This starts with 1152 samples from channel
1 and ends with 1152 samples from channel 7.
Testdata.txt
This 2-D matrix has the dimension 180x8064 and contains 180 trials of test
data. This test data was recorded after the train data (during the
same day) day. The 180 trials belong to either class 0 or class 1.
The matrix has nearly the same structure as the train data matrices exept
for the missing class tag. Thus, every line contains 7 times 1152 samples.
Performance criteria:
Please provide your class rating (0 or 1) for every trial in Testdata.txt.
Please keep the order of trials unchanged when providing your results.
The performance of your classification will be the correct response rate
defined by the number of correctly classified trials divided by the total
number of trials.
Reference:
- Birbaumer, N., Flor, H., Ghanayim, N., Hinterberger, T., Iverson, I., Taub,
E., Kotchoubey, B., Kübler, A., & Perelmouter, J,
A Brain-Controlled Spelling Device for the Completely Paralyzed,
Nature, 398, 297-298.