[Eeglablist] Fwd: Problem using CleanLine

[Makoto Miyakoshi]

Dear Chris,

> noise ranging all the way from 10 up to 14 dB

Oh that's BIG.

First of all, CleanLine can only deal with sinusoidal artifact. I don't
know the source of your 20Hz noise, but you'd better confirm what it is.
You also want to confirm how temporally stable it is. To do this, bandpass
filter the data 18-22 Hz or around and see the time series. Do you see
rather fast wax and wane of it? Compare your CleanLine sliding window
length to the approximate time constant of the wax and wane of the target
noise. If the former is too large, of course CleanLine would fail.

That being said, this seems most suspicious.

> Yet when I save the 'cleaned' file, the spectral peak at 20Hz remains
unchanged (despite the window showing original and cleaned spectra for
selected channels shows clear differences in the frequency band that I'm
interested in cleaning). Any input would be most helpful, all the best.

I wonder if EEG is not updated correctly after computation. Make sure that
you close the windows by pressing 'ok' buttion only... sometimes this
solves the problem.

Just in case why don't you tell me your Matlab version, EEGLAB version, and
OS.

Makoto

On Thu, Mar 31, 2016 at 9:15 AM, Christopher Barkley <barkl025 at umn.edu>
wrote:

> Apologies for the double email, but I wanted to add, to maybe assist in
> resolving this issue, that when the CleanLine script is actually running,
> it states that it is actually is removing *large* amounts of noise in and
> around 20Hz (I have tried different bandwidths, as well as tinkering with
> the step size and window size parameters), noise ranging all the way from
> 10 up to 14 dB. Yet when I save the 'cleaned' file, the spectral peak at
> 20Hz remains unchanged (despite the window showing original and cleaned
> spectra for selected channels shows clear differences in the frequency band
> that I'm interested in cleaning). Any input would be most helpful, all the
> best.
> - Chris
> ---------- Forwarded message ----------
>
> From: Christopher Barkley <barkl025 at umn.edu>
> Date: Thu, Mar 31, 2016 at 2:19 AM
> Subject: Problem using CleanLine
> To: eeglablist at sccn.ucsd.edu
>
>
>
> Hi all,
>      I am having issues using Cleanline (The current release), and I am
> using EEGLAB 13_4_3b), while following (with minor adjustments) Makoto's
> pre-processing pipeline (
> http://sccn.ucsd.edu/wiki/Makoto%27s_preprocessing_pipeline).
>      We have noticed a strange 20Hz artifact in some of subject's data and
> want to use CleanLine to eliminate it rather than a traditional notch
> filter for fear of creating a band hole at this frequency. The issue is
> that, after some messing around with various solutions, I've noticed that
> CleanLine only is effective at reducing noise at this frequency when I
> apply it to epoched data, but not continuous, data, and I was wondering if
> anyone has any idea on why this might be.
>      I plan to remove bad channels using TrimOutlier, interpolate, and
> then reference to the CAR prior to epoching, so I don't want to delay
> referencing, channel removal, etc. until post-epoching for fear of
> introducing all sorts of noise into my data. Any help on this would be
> *much* appreciated, many thanks.
>
> Chris Barkley
> College of Pharmacy
> University of Minnesota
>
>
> _______________________________________________
> 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
>



-- 
Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20160404/b30921c2/attachment.html>