FFT on EEG data (help)

[olivia_90]

Hi!

I made one simple vi for loading an data in .tdms file format and perforimg a FFT on this data (I attached a photo of block diagram). I have a problem with representation of result on EEG Spectrum. It shows some illogical results, which you can see in photos I attached. 

Do you thik there is only a problem with setting of EEG Spectrum for representation or there is a bigger problem with FFT Analysis? Any suggestion or comment will be very helpful.

There are 3 photos in attachment: one of block diagram and two of front panel with different data.

Thank everyone in advance!

[ZJ.Gu · Accepted solution]

Hi,

Does the EEG signals multichannel in your tdms file? From the waveform, it looks yes. Then the Spectrum VI will use the first channel of the array. Please make sure it is correct and also make sure the sampling frequency is correct.

The better way is to insert an index array node to choose which channel you would like to process and feed into the Spectrum VI. You don't need to wire sampling rate if it is included in the waveform.

Thanks!

ZJ Gu

[Haristhota]

Could you suggest me a best technique for detecting real time cognitive state from single channel EEG?

Band pass filtering and FFT, calculating bins of spectrogram over a band of frequencies will give accurate classification?

applying PCA and ICA is better?

I am really confused,

people from literature have been used some advanced classification techniques like nueral networks & support vectors etc., what is their advantage? what are the demerits of spectrogram techique?