AFAIK nobody is working on this

just be careful that filters you use are zero-phase

Another option is to adapt the line filtering routines from Chronux, as they don't require the use of band-stop filters.

heard this one before on the mailing list (probably you mentioned it:)
I should look into it once more.
thanks for the (re)pointer

On 24.10.2012, at 23:54, Eric89GXL notifications@github.com wrote:

Another option is to adapt the line filtering routines from Chronux, as they don't require the use of band-stop filters.

—
Reply to this email directly or view it on GitHub.

Basically it uses multi-taper estimation combined with an F statistic to infer if there were significant sinusoidal components in the data. You can even specifically tell it which frequencies to test, which is nice because you could restrict it to the line freq and its harmonics. I haven't personally tried it on M/EEG data, though...

sounds cool.
would be good however

On 25.10.2012, at 00:07, Eric89GXL notifications@github.com wrote:

Basically it uses multi-taper estimation combined with an F statistic to infer if there were significant sinusoidal components in the data. You can even specifically tell it which frequencies to test, which is nice because you could restrict it to the line freq and its harmonics. I haven't personally tried it on M/EEG data, though...

—
Reply to this email directly or view it on GitHub.

woops, sorry, something was missing--

would be good to have one or two references from meeg studies using this technique.

but anyways, sounds very nice.
let me look into this soon.

D

On 25.10.2012, at 00:14, "Denis A. Engemann" denis.engemann@gmail.com wrote:

sounds cool.
would be good however

On 25.10.2012, at 00:07, Eric89GXL notifications@github.com wrote:

Basically it uses multi-taper estimation combined with an F statistic to infer if there were significant sinusoidal components in the data. You can even specifically tell it which frequencies to test, which is nice because you could restrict it to the line freq and its harmonics. I haven't personally tried it on M/EEG data, though...

—
Reply to this email directly or view it on GitHub.

Just a reminder to @agramfort to look into notch filtering for 0.6, as our code could easily knock out a small band of frequencies.

also see #98

On 10.12.2012, at 20:48, Eric89GXL notifications@github.com wrote:

Just a reminder to @agramfort to look into notch filtering for 0.6, as our code could easily knock out a small band of frequencies.

—
Reply to this email directly or view it on GitHub.

@agramfort do you have input on this? I assume you haven't had a chance to look through other packages...

Since I can't work on #11 while traveling, I might try to work on a few options for this: an FIR/FFT band-stop, an IIR band-stop, and I'll try doing what is done in chronux (although I've read through some of that code before, and it's not the easiest thing to understand). Let me know if you have other methods ideas.

indeed no time to look carefully. I would fly with the code of the standard
toolbox in the domain for inspiration...

We are using a perl translation of this matlab window-sinc filter,
which does a very nice job for removing power line artifacts from MEG data.

http://www.gomatlab.de/window-sinc-filter-t19156.html

I always wanted to translate it, but always other stuff seemed more
important.

On Fri, Dec 21, 2012 at 9:55 AM, Alexandre Gramfort <
notifications@github.com> wrote:

indeed no time to look carefully. I would fly with the code of the standard
toolbox in the domain for inspiration...

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/98#issuecomment-11606594.

Closing this to move discussion to #316.