[Eeglablist] Eye Blink Sampling Rate

[Emmanuelle Tognoli]

Dear Marwa,

I assume that when you talk about blink, you are interested in the simpler
question of the artifact itself (the record of a blink also involves
movement of eyelids and eyes and associated motor, oculomotor and
somatosensory neural activity, a complex problem).

Regarding your questions 3 and 2, two thoughts come to mind:

---3-What  it means for the blink to have 8-13 Hz rhythm then?---

You will find a modest but non-zero energy at this frequency range that
pertains to the blink artifact, not a "rhythm" per se, but contribution to
this frequency band nonetheless. The blink artifact has a very odd shape
from the standpoint of Fourier and related techniques on which most of the
spectral analyses are based. A blink artifact is nowhere near sinusoidal,
it has asymmetries both vertical and horizontal. To represent this odd
shape, more frequencies are needed, and this is why your blink leaks
energy in the 8-13Hz band.

---2-What is the sampling rate that I should use for eye blink?---

If you analyze both the blinks and some EEG together, it seems impractical
to have two sampling rates in your study. Unless you meet resource
limitations for an analysis, you might want to keep the 250Hz sampling
rate. With this sampling frequency, you have a nice representation of the
odd shape of your artifact. No matter what, if required to downsample, the
20Hz seems a bit low, even if most of the blink energy will be in the
range of [DC-5Hz]; again because of the blink shape. Especially, the
lowpass filter that would be required prior to downsampling would do more
harm than good at such proximity from frequencies of interest and with
such an asymmetrical signal. It might be better to avoid this altogether.

---And finally, you asked about the convenient number of samples that can
represent the blink.

This varies from subject to subject and to a lesser extent, from blink to
blink (variance especially if subjects are trying not to blink: as they
volitionally reopen their eyelids, a blink might be more or less
shortened). As a rule of thumbs, the blink artifact might last 1/4th to
1/2 sec: then most of its signal would concentrate over 63-125 samples (at
250Hz), or 5-10 samples (at 20Hz).

With kind regards,
_________________________________________________

Emmanuelle Tognoli - PhD
Center for Complex Systems and Brain Sciences
Florida Atlantic University
http://www.ccs.fau.edu/~tognoli

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.