Currently, the Fourier analysis we are performing projects the signal on many cosine waves at different frequencies to find the frequencies present in the signal. However, many of the common EEG waves that appear are not cosine waves.
To increase the accuracy and confidence in our algorithm, an interesting future development would be to perform Fourier analysis with common seizure EEG waves as opposed to cosine waves. In doing this, we would be able to sweep across the waveforms that you would expect to see at various frequencies in order to have a more accurate projection.
As an example, this could be done with a wavelet transform. Some resources to information on wavelet transforms can be found below.
Suggestions? Potential Applications? Drop a note!