[Eeglablist] How to deal with 1/f noise in the low frequency oscillations for on-line experiments?

Gangadhar Garipelli gangadhar.garipelli at epfl.ch
Thu Nov 11 04:40:51 PST 2010


Hello Philip & eeglablist,

Thank you for your prompt response. Yes, you are correct! I should apply
high pass filtering to remove a variety of neuronal and non-neuronal
artifacts in the very slow frequency range[I, II]. Eventually, a high
pass of 0.1Hz cut-off is good enough. But the problem is, I need to have
a sharp transition for the filter and should have "almost" no
group/phase delay for the on-line experiment.

"I quote again that I need to implement such a filter on-line and almost
in real-time wherein I can afford a time delay (resulting from filter's
group delay) of ~1s."

If I manage to have such a filter, I would not loose any signals in the
range [0.2 0.6] & [0.6 0.8]Hz that are relevant to my experiment.

I am expecting various suggestions, using either signal processing or
machine learning approaches or combination of both. I had suggestions
already, using matched filters, lock-in amplification, amplitude
modulation using choppers, exploiting spatial information etc. But none
of them are satisfactory as of now.

If one of you dealt with such a problem already, I would be very
grateful to learn from your experience! :-)

References :
[I] Vanhatalo S, Viopio J, Kaila K. Full-band EEG (FbEEG): An emerging
standard in electroencephalography. Clin. Neurophysiol., 116(1):1-8, 2005.

[II] Vanhatalo et al., 2004b S. Vanhatalo, J. Voipio and K. Kaila,
Infraslow EEG activity In: E. Niedermeyer and F. Lopes da Silva,
Editors, Electroencephalography: basic principles, clinical
applications, and related fields, Lippincott-Williams & Wilkins, Boston,
MA (2004).

Sincerly
Ganga

On 11/11/2010 05:16 AM, PHILIP GRIEVE wrote:
> not sure what noise source you are referring to - do you mean the noise from the DC coupled EEG amplifier?  i would think that the real problem in discerning these very slow cortical signals is the electrode artifact voltages generated at the contact of the skin and the electrode - these "half-cell" potentials drift around and can be very large as the potentials from the two EEG electrodes oppose each other and their difference cause artifact at the mm volt level - a high pass filter can remove this artifacts but also will remove your signal! 
> 
> 
> On Nov 5, 2010, at 7:50 AM, Gangadhar Garipelli wrote:
> 
>> Dear all,
>>
>> I work with low frequency oscillations of the brain while a human
>> subject is cognitively engaged in a task. From the off-line analysis
>> (using zero-phase band pass FIR filters on full-band EEG), I discovered
>> that task-related cognitive signals are located in the range of [0.2
>> 0.3]Hz and in [0.6 0.8]Hz. The fluctuations/oscillations ( formally
>> called very low frequency oscillations VLFO, or infra slow oscillations
>> ISO) below 0.2Hz are REAL devil due to 1/f nature. The noise power is
>>> 100 times higher than signal's power.
>>
>> Now as per my experimental demands, I need to estimate on-line in
>> real-time the signals mentioned in the above range and manipulate
>> stimulus presentation. Ideally, this eventually means I need to have a
>> very sharp high pass filter with almost zero group/phase delay. Which
>> sounds impossible!
>>
>> However, I should come up with a decent trade-off between SNR and
>> phase-delay. Do you have any suggestions? All suggestions ranging from
>> signal processing/machine-learning to hardware to solve this problem are
>> most welcome!
>>
>> Thanks in advance!
>> Sincerely,
>> -- 
>> Gangadhar GARIPELLI,
>> Doctoral student,
>> EPFL, Switzerland
>>
>> _______________________________________________
>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu
>> For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu
>>
> 


More information about the eeglablist mailing list