I am investigating the oscillatory brain activity (EEG 32 channels) of non-clinical adults during a Virtual Reality task.
During this task, participants can move their head in order to inspect the virtual environment.
I am using EEGLAB toolbox for the analysis.
I will use ICA in order to reject some kind of artifacts ( blink, eye movements, muscle artifacts).
How can I control artifacts of movement (e.g. head movements) ?
Thanks
For a successful analysis of EEG or MEG signals, “clean” data is required. That means that you should try to reduce variance in the data due to factors unrelated to your experimental conditions. One of the factors that is difficult to control for is the presence of artifacts in the data. These artifacts can be physiological or the result of the acquisition electronics. The strongest physiological artifacts stem from eye blinks, eye movements and head movements. Muscle artifacts from swallowing and neck contraction can be a problem as well. An example of artifacts related to the electronics are ‘SQUID jumps’ or spikes.
Of course it is best to try to avoid those artifacts in the first place. For instance, you might give your test-subjects some well-defined time interval between the trials in which they are allowed to blink, but ask them to withold blinks in the time intervals of interest. Sooner or later, however, you would like to go through the data and make sure artifacts are detected, removed or corrected.
Automatic artifact rejection in FieldTrip is a sophisticated and complicated approach to artifact rejection, that without full understanding of all the steps involved will unavoidably lead to more harm than good. Only when you fully understand your data and the artifacts you will be dealing with, will you be able to set appropriate parameters for automatic artifact rejection
The following steps are used to detect artifacts in FieldTrip’s automatic artifact rejection (see figure below
This procedure will result in an artifact definition, a two column array with the onset and offset sample number of every detected artifact. This ‘artifact definition’ (‘‘.artfctdef’’) can then, in a separate step, be used to reject (parts of) trials in your trialdefinition (so before reading the data in using ft_preprocessing for your subsequent analysis), or rejecting (parts of) data already in memory. Both methods use ft_rejectartifact
EEG is a mechanical recording of the local skin micro-displacements with the help of electro-chemical cell involving human body as a factor of modulation of galvanic potential of the cell. So, it is, in principle, impossible in a case you cannot define exactly what is a signal and what is an artifact. Still, when you are able to record "separately" an artifacts (e.g. using a capacitive 'non-contact' electrodes) with adaptive filtering it is possible to differentiate at some degree the particular components of the signal (referred as EEG local activity).
Do you have any hypothesis in advance? This would determine what aspect of the measured EEG is of interest - event related EEG signatures or variations of spectral content of the signal and the respective aspects of localization of such effects.
Could be that rigorous band pass filtering would help during preprocessing of the data.
- Badges
- Science topic