what should be the minimum sampling frequency required for EEG?

Discussion in 'Programmer's Corner' started by mayank rob, Oct 21, 2012.

  1. mayank rob

    Thread Starter New Member

    Sep 18, 2012
    what should be the minimum sampling frequency required for EEG if want to sucessfully detect the drowsiness level of any person. The ready made products which i found was
    EEG EPOC machine but its sapling frequency is approx. 128Hz,which i think is very less, but since i saw many projects implemented succesfully using this machine,can i use it for my project?
  2. paulktreg

    Distinguished Member

    Jun 2, 2008
    I have only come across GE Medical Anaesthesia Patient monitors and they sample at 100Hz with their E-EEG module here.
  3. wmodavis

    Well-Known Member

    Oct 23, 2010
    You need to know the characteristic of that part of the waveform that is an indicator of phenomenon that you want to detect. Then you pick the sampling frequency. So do your homework. Look into the frequency spectrum if the applicable indicator. Or assume and copy.

    "The problem with copying is it skips understanding and understanding is how you grow. You have to understand why something works."
  4. t06afre

    AAC Fanatic!

    May 11, 2009
  5. mayank rob

    Thread Starter New Member

    Sep 18, 2012
    well you have found the exact IEEE paper that i am referring to :p
    well they are sampling it at rate of 500Hz and then down-sampling it to 250 Hz.

    would you please go through Emotiv EEG EPOC,its internal sampling rate is approx. 2kHz,and it downsamples it to 128Hz and sends it to PC via bluetooth to the computer for analysis.
    so i just wanted to make sure that sampling rate is quite enough do get any meaning information so that signal processing can be done.
  6. SPQR


    Nov 4, 2011
    If you look at a similar biologic signal, the ECG, its relevant frequency information is higher than that of the EEG. An ECG sampler must be able to detect a pacemaker spike, thus you can see sampling rates at high as 3kHz.

    The EEG has much lower relevant frequencies, most of the spectral energy is found at less than 30 Hz, thus applying Nyquist, 60 Hz would capture most of it.

    But seizures can have frequencies as high as 100Hz, so 200Hz sampling would be reasonable. But given the cost of the hardware, memory and software, oversampling to 0.5-1kHz doesn't seem unreasonable.