CS295J/Project Schedule:HG-DT/Interview JimCav EEGDataAnalysis

Q: Could you please describe the experiment from which the EEG data was obtained?

A: The experiment we performed is designed to study if human's color perception and shape perception are associated with different brain regions. During the experiment, each subject is showed several pictures of different objects consecutively, and is asked to report either the shape or the color of the object. Their brain activity is recorded through the whole process.

Q: Is there any pre-processing step before the actual analysis of the data?

A: Yes. For the raw data obtained, I will first perform a quality check manually. I will look at the EEG signals and try to identify and discard signals that displays unusual instability or discontinuities. Signals with these characteristics are most likely to be noises.

Q: What tool do you use to analyze and visualize EEG data?

A: I write custom Matlab code with a Matlab Toolbox called EEGLAB (link: http://sccn.ucsd.edu/eeglab/) to process the data.

Q: What is the general procedure for you to analyze EEG data?

A: When I am analyzing the EEG data, I work with three types of visualization: temporal, spatial, and power/phase. I need to go back and forth between temporal and spatial representations, and if I find something interesting, I will go further to examine the corresponding power/phase representation. What I usually do is, I select one or two epochs according to my knowledge about the experiment condition - usually one when the subject is identifying shape and the other when the subject is identifying color - and compare the two visualization to see if there is anything interesting about the activation pattern. If not, I will select some other epochs and regenerating the visualization. Sometimes I also want to look at the time-varying signal at a particular electrode in one epoch. If I don't find anything interesting at a particular electrode, I will switch to another electrode.