COMP6620 Signal Analysis for Computing

SSVEP is the response of the brain when visualising flashing objects and recorded using electroencephalogram (EEG) in the visual cortex. SSVEP response will resonate at the same frequency of the flashing object that the user gazes upon. So, if a block flashes at 10 Hz, the SSVEP response would also have a frequency component at 10 Hz. Here, the data is from one user who looked at four objects each flashing with a fixed integer frequency (which you will have to determine) in the range of 8 to 30 Hz. The time when user gazed at each object varied from 1, 2 or 3 seconds. User also occasionally took a break (i.e. non-gazing) for either 1, 2 or 3 seconds where EEG only recorded random noise. The total length of the EEG data varies from 20 to 30 seconds (in blocks of 1 second) and the sampling frequency used was 250 Hz.

 

SSVEP or non-gazing segments do not overlap, i.e. either the user looked at a specific block or it was a non-gazing period.

The data is also corrupted by two noise:

1. Baseline sinusoidal with frequency lower than SSVEP

2. Powerline (50 Hz) interference

Your task is to segment and label corresponding segments with either one of the determined SSVEP frequencies or non-gazing as shown in
Table 1 below (just as an example, data provided to you will be different). Furthermore, in your report, you will also need to give details on
the approaches utilised (and the justifications) with Matlab code snippets and figures as appropriate. The full set of Matlab codes (as one file)
should be uploaded to the raptor folder.

1. Load the signal and decide the length of the signal (2 marks)

2. Design a notch filter to remove 50 Hz powerline interference (3 marks)

3. Design a high pass filter with appropriate order and cut-off to remove the baseline noise (4 marks)

4. Segmentation of the signal to 1 second lengths each (3 marks)

5. Design appropriate filter for the SSVEP data (4 marks