Analysis of EEG Data Using an Adaptive Periodogram Technique

We report a time-frequency analysis method for EEG data processing using an adaptive periodogram technique (APT). The time duration of a coherent signal is estimated using a time window that adapts its length and position to locally maximize the periodogram. Since this technique determines a signal´s coherent length, it achieves optimal resolution in both the time and frequency domain. We apply this technique to analyze the EEG data of two seizure events. During the seizures, the dominant frequency evolved continuously from about 8 Hz to 1 Hz. Before the seizures, there existed a weak signal at 8 Hz, which was clearly manifested in the coherent time. These preliminary results suggest that APT is a useful EEG analysis tool.