BCI-walls
These scripts calculate SNR and the SNR-walls of EEG
Dataset
Download the dataset from https://researchdata.gla.ac.uk/1258/ and unpack the zip outwidth of this repository so that the directory structure looks like this:
--+--bci_wall_analysis
|
+--gla_researchdata_1258
Pre-requisites
The following python modules are required:
- matplotlib
- numpy
- scipy
Commandline parameters
All python commandline tools have the same parameters:
- -p participant number between 1 and 20 (default is 20)
- -s The second at which the data analysis starts (default is 60sec)
- -t The task which can be: "jawclench", "read", "colour", "wordsearch", "sudoku", "phoneApp", "lyingEC", "lyingEO"
- -h Help
All parameters are optional and have default values set.
explore.py
Plots the data in the timedomain and frequency domain of one subject and for all tasks (or one task if specified).
snr.py
Calculates the signal to noise ratio from one subject doing one task. It can be used both as a module and main program.
p300.py
Plots the signal power of the P300 peak which is used in the SNR calculations.
noise_wall.py
Calculates the SNR-wall. It can be used both as a module and main program.
bci_wall_one_subj_analysis.py
Calculates the SNR-walls and SNRs for one subject.
bci_walls_with_t_test.py
Calculates all SNR and SNR-wall values for all task over all subjects and performs t-tests if it is significantly possible to detect conscious EEG changes.
Pre-set parameters for the frequency bands:
- -f: full range
- -m: 8-18 Hz (used for BCI for motor imagination)
- -n: 8-12 Hz (alpha frequency range used for BCI)
- -d: derivative (1st order highpass used for BCI)
- -e: 0.1-3 Hz (eyeblink frequency range)
Alternatively, set the frequency range yourself:
- -a [minF] -b [maxF] frequency band when detecting bandlimited events (default: full spectrum)
do_all_stats.sh
Runs five bci_walls_with_t_test.py scripts for all pre-set parameters in separate
processes.
Credits
Bernd Porr
Lucía Muñoz Bohollo