Averaging time-frequency analysis

Hi,
I want to study the dynamic of Gamma Power prior to and during seizures. More specifically, I want to segment the seizures into 3 equal epochs. then for each epoch, I want to calculate the Gamma Power and finally, I want to compare the Gamma power between these three ictal epochs as well as pre-ictal epochs.
Should I compute the FFTs or use time-frequency decomposition (Morlet transform) and average it over each epoch or you suggest any other approach?
I greatly appreciate it if you respond in detail as I am new to these calculations.
Thank you

If you are interested to obtain only one single value for each of your epoch, it would be much faster and easier to use a FFT-based approach rather than Morlet wavelets.
If you epochs are not too short (more than a few hundreds of ms), then you would obtain more stable results by using the process " Frequency > Power spectrum density (Welch)" rather than a simple FFT:
https://neuroimage.usc.edu/brainstorm/Tutorials/ArtifactsFilter#Evaluation_of_the_noise_level