nt_zapline_plus: nremove checked on wrong dimension

[0xlevel]

In v1.2.1 nt_zapline_plus, line 54, if nremove>=size(x,1); error('!'); end checks the wrong dimension. I think, it should be size(x,2).

I encountered this when trying to use zapline plus to clean two channels, only. An exception will be thrown later (in line 146), but I thought, you might want to know. The error can be bypassed by using fixed error components (= 1), but can perhaps be addressed, when set to adaptive. It is also possible, that I missed something and in that case I would be glad for advice. Cheers!

[MariusKlug]

I can confirm this, I think you are right. I recommend setting this to the second dimension and using adaptive cleaning instead of fixed cleaning with 1 dimension.

Although I do think that using zapline-plus for two channels only will likely to be problematic as spatial filtering requires at least some precision there. Where does your EEG data come from?

[0xlevel]

Hi, thank you for your reply. I agree, that the 2 ch setup was indeed a very special case (see below). ;) I wanted to quickly pass on my observation, though, as adaptive cleaning also failed in my case. My use case was somewhat abductive, to put it mildly: I was briefly looking into methods to remove power line interference from EMG(!) data. A notch around 50 Hz (in my case) would already destroy too much information.

[MariusKlug]

Yeah it is a valid bug report, so thanks :)

I see. Hmm, maybe in your case the cleanline function of EEGLAB might be more useful, or did zapline in the end work for you?

[0xlevel]

You are welcome. :) Yes, cleanline works fine, but has trouble working with non-stationary noise. I also looked into PREP’s cleanLineNoise, which seems a bit excessive and the implementation of the Regression-Subtraction method (Mewett, 2001) by Eiber & Pietersen, which also works, but would require some additional effort. So, yeah, Zapline plus is still tempting, but perhaps later™. Cheers!