I am analyzing EEG signal and I am unable to understand the term Time-Frequency Power Spectrogram. Please Can anyone help me?
Thank you in advance.
$\begingroup$
$\endgroup$
Add a comment
|
$\begingroup$
$\endgroup$
2
Some signals (those stationary enough) can better explained, or described, through a Fourier transform. Since it is energy preserving, one usually uses the power spectral density, or "the squared Fourier spectrum".
EEGs can be stationary over local time frames. Hence, time-Frequency power spectrograms are just power spectral densities computed on sliding overlapping windows.
answered Feb 9 '19 at 18:08
-
$\begingroup$ Ya I found it to be Power Spectral Densities, Thanks a lot sir $\endgroup$ – padmapriya vijayakumaran Feb 12 '19 at 2:54
-
$\begingroup$ And would you evaluate the answers given? $\endgroup$ – Laurent Duval Feb 13 '19 at 19:49
$\begingroup$
$\endgroup$
1
Short-time Fourier transform
Traditional Fourier analysis provides no information about evolution of frequency components over time. However, Short-time Fourier transform (STFT), a method which applies a short-time window to the signal and performs a series of Fourier transforms within this window as it slides across all the data, can overcome this limitation, providing a time–frequency representation of the signal.
In MATLAB, try
doc spectrogram
Continuous Wavelet Transform
Alternatively, continuous wavelet transforms (CWT) provide a useful approach in investigating non-stationary signals, this is often regarded as an “optimal” solution with regard to time and frequency resolution. Reportedly, CWT perform better at detecting abrupt changes in non-stationary signals compared to STFT.
In MATLAB, try
doc cwt
Generally, EEG signals have a low signal to noise ratio (SNR), therefore it is preferred to use ensemble averaging over trials for your time-frequency analysis (TFA), should the study design allow it.
-
$\begingroup$ Thanks a lot for the explanation. It was halpful $\endgroup$ – padmapriya vijayakumaran Feb 12 '19 at 2:53
