OpenViBE forum

Hello,

I am using EEGLAB to conduct analysis on the "Motor Imagery of Hands" dataset that is found in the downloads section of this website (Downloads > Datasets > Dataset #1 - Motor Imagery of Hands

In EEGLAB, I am trying to extract data epochs from the data provided. EEGLAB requires you to input epoch limits (start, end) in seconds for the event type (which would be "Left" or "Right").

I read the reference paper given with the dataset (Motor Imagery and Direct Brain–Computer Communication) and from what I understand the arrow pointed left or right from 3000ms to 4250ms. However between each trial there was a randomized interval from 500ms to 2500ms so how should I go about finding epochs of this dataset?

Thank You

Hi,

you can investigate the markers (stimulations) in the signal, look from the ov web documentation which numbers correspond to which triggers, and then in Matlab compute the durations between some start marker and marker corresponding to the trial. I'm not sure if EEGLAB automatically supports variable length trials, you might have to hack it or just use the shortest duration that covers all the trials.

Cheers,
Jussi

Hello,

Thank you for your response! I noticed the "Labels" file has stimulation codes which occurred during the experiment. I cannot find a document showing what stimulation codes the numbers in the "Identifier" column are on the "OV Stimulation Codes" page (http://openvibe.inria.fr/stimulation-codes/)

Thank You!

Hello,

I did some digging and realized the epochs last 0.0625s in duration (total number of epochs is 9931 for subject 1).

I am trying to extract epochs from the data (https://sccn.ucsd.edu/wiki/Chapter_05:_ ... ata_Epochs) but keep getting an error stating "Remove mean of each data channel". Any Idea what this is being caused by?

Thank You!

The stimulation codes on the webpage are unfortunately hexadecimal, you need to convert them to the decimal notation to match. Usually the codes in .ov files belong to that set, but its also possible to use some custom numbers.

Regarding EEGLAB, it might be better to ask the EEGLAB mailing list or forum, if they have.


Cheers,
Jussi