EEG Channel Data Filtering

I've been assigned a task to filter raw data from an EEG channel. I haven't been told what filter to use, I've only been given the input data and a picture of what the output should be:

Raw Data:

Filtered Data:

I'm a beginner to signal processing in MATLAB so any relevant guidance would be greatly appreciated. Thanks.

Answer 1

[Star Strider]

The first step would seem to me to calculate the fft of the raw data to see what the frequency content is.

Since neither the filter nor apparently any of its properties were stated, use the plot of the fft results to determine the filter characteristics. Since there appears to be significant baseline variation and possible baseline offset, calculate the fft on the signal after removing the mean of the signal (the d-c component) so that the remaining peaks are more apparent. The filter itself will have to remove the d-c offset and baseline drift, as well as high-frequency noise, so base the filter design on those criteria. Use the fft results to select the frequencies.

If the noise is broadband — present in all frequencies (although that does not appear to be apparent in the image of the filtered signal) — no frequency-selective filter will remove it, requiring other techniques such as wavelets or the Savitzky-Golay filter.

This sounds like a fun (if labour-intensive) project!

.