Preproc - thresholding negative z-scores too?

[jan-mathijs schoffelen]

Dear Alex,

The answer to you question would be in this case: it's a historical
thing. The Z-value thresholding was conceived to be done on filtered-
and-subsequently-rectified/hilberted data. No need to look for low
Zvalues in that case. In itself, there's no reason not to build in an
option to look for negative Z-values as well. I would gladly help to
build it into the release version for you. I would think of an
option: cfg.artfctdef.zvalue.tail, which will be 1 by default (that
is: only looking at high positive deviations), but could be -1 (for
the large negative ones), or 0 (for both sided). Could you propose an
adjusted version of artifact_zvalue to me, and I'll check it and
build it in.
As for the reduced discriminability: I still don't understand why
this is the case. I assume you have more than 1 EOG channel?
Probably, one of these guys is picking up most of the artifacts, but
if you use both channels in a single run of artifact_eog, the same
threshold is applied to both channels, which indeed would affect the
sensitivity. An alternative would be to call artifact_eog twice, with
the different EOG in cfg.artfctdef.eog.channel, so that you can use
channel-specific thresholds.

Yours,

Jan-Mathijs

On Oct 28, 2008, at 1:14 PM, Alexander Maye wrote:

> Dear Jan-Mathijs and All!
>
> Thank you for the quick response. Hilbert indeed flips my eye
> blinks towards positive
> z-values - at the cost of reduced discriminability. That is to say
> that the amplitude of the
> z-score of the eye blink is twice as high without Hilbert (but in
> the "wrong" direction). Apart
> from that I still would like to know the rationale behind
> considering deviations of the z-value
> towards +Inf as artifacts, while those towards -Inf not. This is
> the same concern as in
> #Item001386 of this list exactly one year ago. Any opinions?
>
> Best,
>
>                          ALEX.
>
> ----------------------------------
> The aim of this list is to facilitate the discussion between users
> of the FieldTrip  toolbox, to share experiences and to discuss new
> ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/
> archives/fieldtrip.html and http://www.ru.nl/fcdonders/fieldtrip.

----------------------------------
The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/fcdonders/fieldtrip.