This is a repository with MATLAB scripts that extract and analyze information from EEG files from a Hero Bitbrain EEG device. This data was collected in a series of experiments develos in the FENNSI investigation group in the National Hospital of Paraplegics in Toledo, Spain. In the experiments, a subject had to a series of tasks: 2 minutes with his eyes open trying not to think anything, 2 minutes doing the same but with the eyes close, 10 minutes of motor imaginary task (thinking about moving your hand without actually doing it) and lastly another 2 minutes wit the eyes open. During the imaginary task, the patient had to imaging a movement with the left or the right hand, depending on the direction of the arrow that showed the software used in the experiment. This experiment was made with people that suffered neuropathic pain from a spinal cord injury. The final goal was to find a relation between the pain and the cortex activity captured with the EEG device.
The files have to be ran in a specific order:
- bandpass_asr.m
- all_excels_afterasr_withoutIT.m
- trials.m
- calc_psd.m
The first one, bandpass_asr.m, loads the data from the files provided by the EEG device and filter the signals. First, a bandpass filter is used, and then the ASR method. It returns the data filtered in csv format. The second file calculates the PSD from every channel in the data and extracts the power of each of the five frequency bands of activity recorded in the brain (alpha, beta, gamma, delta, theta). The third file, trials.m, calculates the PSD from the trials in the motor imaginary task. Each trial was marked by the arrow that showed in the program. The last file, calc_psd.m, extracts the power of the frequency bands of the PSD of the motor task. First, it calculates the mean of all the PSD of the trials of each hand, and then it extracts the power from that signals.
With the processed EEG files, the new script "test_ERPanalysis_ERDERS_spectrogram.m" does the following steps:
- Compute ERP of the trials of each experiment
- Compute ERD-ERS ratio of the ERP and show the spectrogram
In the script, the following steps are observed:
First, the signal power is calculated. Then, the trials are separated (2 seconds of baseline and 3 seconds of motor imagery) and averaged, generating the ERP (one per channel and separated by right hand and left hand). The ERP is saved in an Excel file named "ERP.xls". The trials for the right hand and left hand will be saved separately but within the same Excel file.
After obtaining the ERP, the signal is smoothed using a moving average window. With the smoothed signal, the ERD/ERS of the signal is calculated. First, the part before the trial (baseline) and the part after the stimulus (post-stimulus) are separated. Then, the following formula is applied:
The MATLAB function "erp_bin.m" computes the ERP of the experiments.
In the ratio, it should be noted that positive values correspond to synchronization (ERS) and negative values correspond to desynchronization (ERD). There should be greater desynchronization in the area where cortical activation is expected. For example, with LEFT motor imagery, the contralateral side, the right hemisphere, is activated, so ERD should be seen on that side (more negative values).
To finish, the spectrogram and the PSD of this ratio are calculated to have numerical values for analysis.
In the end, it should generate two Excel files in xlsx format (one for the left and one for the right), where each sheet represents a channel, and within each sheet, the PSDs are separated by experiments (129x9 matrices). These matrices contain the spectrogram data (time-frequency) for the ERD/ERS ratio.
In some cases, the matrix will be filled with zeros. This happens because the signal has some issue, and the ERPs are not calculated correctly, leaving an empty signal. When calculating the ERD/ERS with these signals, an error occurred because it was dividing by zero. I fixed it so that, if this happens, it writes a matrix filled with zeros.