I am trying to compare EEG features such as power spectrum and standard deviation to detect emotions. I epoched the data and saved in .set format. Highly appreciate your support to continue the rest of the project.
Yau may look to the channel properties of intersted channels of yours also you can look to the components by running ICA
Ahamed, What is the toolkit I should use to extract features? should I use EEGLAB or BCILAB?
Dear Ahamed, a fairly straightforward manner to obtain features from an epoched EEG is to compute the frequency spectra (FFT), relative to your experimental conditions (valence: neutral, positive & negative, for instance). Then, you might be interested to compare between conditions, and find differences.
Alternatively you can also compute time-frequency spectra to each condition.
All these computations may be done using EEGLAB.
For Feature extraction u have to use a suitable feature extraction methodology according to ur application such as PSD, FFT, wavelet and others and after you may need to use features reduction techniques to reduce the features vector then a suitable classifier can be used to classify the signals according to the application also.
The EEG is a very complex signal. Therefore, in addition to frequency analysis, you may be interested in extracting complexity features (e.g. fractal dimension or entropy).
As for me, I will see the future EEG in clinical analytical approach for the understanding the process of brain injury and brain disfunction. http://angio-veritas.com/en/technologies/
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Waves