 |
 |
 |
 |
Much of our current work makes use of wavelet analysis
The figure above shows three example Gabor wavelet functions for progressively higher frequencies. These functions illustrate the way in which a signal is analyzed, point by point, with the greatest weighting given to the sample at a given time, and progressively lower weightings given to the signal results before and after that time. The signal is sampled progressively, sample by sample. Wavelet analysis makes it possible to decompose signals that contain a number of frequencies, have rapidly changing components, and contain non-stationary elements. We combined wavelet analysis and cross-correlation analysis. This has the advantage that our methods can reveal not only transient EEG changes within discrete frequency ranges but also correlations between sites (i.e. as a measure of "synchronization"), allowing us to optimize the resolution capabilities of the method in time, frequency and space. (Correlation is a time-domain analog to synchrony in the frequency domain.) Wavelet cross-correlation coefficient (WCC) analysis enables us to determine the degree of waveform similarity between two different time series. It provides a quantitative measure of relatedness of two signals, usually from different recording sites, as they are progressively shifted in time with respect to each other. It also can reveal common components that occur at same moment in time, or at a constant delay. We have found that WCC values trend lower under circumstances during which afterdischarges are more quickly terminated by brief pulse stimulation.
Figure: We assessed WCC values under four conditions. WCC values trend lower under conditions when ADs stop more quickly.
1A shows results for afterdischarges (ADs) that stopped within 2 seconds.
1B shows results for ADs that stopped in 2-5 seconds.
1C shows results for ADs that stopped in more than 5 seconds.
1D shows results for trials when no ADs occurred.