<div dir="ltr">This is an interesting question!<div><br></div><div>I think PCA is better than ICA in eye-blink removal. Because what ICA does is first ESTIMATING the eye blink component and then linearly removing that component from the data set. However, in PCA approach, a better estimation of eyeblink component is employed (using for example eye-blink system) to linearly remove that component from the data set---note that this component also does not contain brain activity. In the cases when we do not have the direct eye-blink measurement ICA is better.</div>
<div><br></div><div>Other ideas?</div><div><br></div><div>Thanks</div><div><br></div><div><br></div><div><br></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On 10 October 2013 10:50, Alik Widge <span dir="ltr"><<a href="mailto:alik.widge@gmail.com" target="_blank">alik.widge@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div><div>Thank you for posting this, as I'm in the middle of this same processing step, and I've been pondering methods as well while following the tutorials. (In my case, 60-channel EEG plus one bipolar EOG in a diagonal configuration.) Your response to Craig contains something that's been bugging me a little bit:<br>
<br></div>When doing the ICA on a test recording, I find (and it sounds like he finds as well) that there is not a single component that captures eyeblinks well (and I have tried various adjustments, such as altering number of components via pre-PCA, clipping or not clipping out epochs that look to me to have substantial slow eye-roll, turning runica.extended on/off). I do get components that have the classic "pair of eyeglasses" or "single eyeball" look... but I get four or five of them on a 58-component decomposition, and that's before we talk about the components that are almost pure 60-cycle noise or temporal EMG.<br>
<br></div>You seem to be telling us, in your comment about orthogonality and percentage of variance captured, that this is actually a *good* thing, because it reduces the chance of removing activity from the frontal pole. Can you help me understand that a bit better? I've felt very nervous about the sheer number of components I'm removing; it feels as though I'm killing a big chunk of the dataset, and doing so somewhat blindly.<br>
<br><div><br></div><div class="gmail_extra">Thanks,<br><br clear="all"></div><div class="gmail_extra"><div><div dir="ltr">Alik Widge, MD, PhD<br>Massachusetts General Hospital<br>Charlestown, MA, USA<br><a href="mailto:alik.widge@gmail.com" target="_blank">alik.widge@gmail.com</a><br>
(206) 866-5435<br><br></div></div><div><div class="h5">
<br><br><div class="gmail_quote">On Thu, Oct 10, 2013 at 4:43 AM, Robert Oostenveld <span dir="ltr"><<a href="mailto:r.oostenveld@donders.ru.nl" target="_blank">r.oostenveld@donders.ru.nl</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi Craig,<br>
<br>
Let me forward this to the email discussion list.<br>
<br>
<br>
On 9 Oct 2013, at 23:27, CR wrote:<br>
<br>
> Hi Robert,<br>
> I wanted to see what your thoughts were on the merits of 2 different methods of removing blinks. I have a 12 minute resting state segment of data, so it has required me to do some things a little differently.<br>
><br>
> Method 1: ICA<br>
><br>
> I break the 12 minute segment into 2 second intervals, since doing ICA on the whole segment gave a poor result.<br>
<br>
why does it give you a poor result? Has the subject been moving? Is there something else that makes the data not compatible with the stationary mixing assumption?<br>
<br>
Or is it the difference in the preprocessing? 12 minutes of data represented in one segment can have drift, whereas 12 minutes of data represented in 2 second snippets will not have the drift (assuming you use the default cfg.demean=yes). Doing a low-pass filter on the continuous data would have a similar effect as segmenting it and demeaning the 2 sec snippets.<br>
<br>
> I apply the resulting unmixing matrix to the 12 minute segment and correlate each component with the EOG to find the most relevant components, and reject these based on a threshold.<br>
<br>
so a bit like<br>
<a href="http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_ica_to_remove_eog_artifacts" target="_blank">http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_ica_to_remove_eog_artifacts</a><br>
with the correlation method of<br>
<a href="http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_ica_to_remove_ecg_artifacts" target="_blank">http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_ica_to_remove_ecg_artifacts</a><br>
<br>
<br>
> Method 2: PCA<br>
><br>
> I do a timelock analysis based on the blink onset point returned by the eyelink system. I then PCA the resulting blink ERF. I then reject the component(s) that account for say 98% of the total variance.<br>
><br>
> Obviously option 2 is much faster. What do you see as the relative merits/problems with the techniques? Technique 1 is largely what the FT tutorials suggest, so what about method 2?<br>
<br>
Option 2 makes the large variance component orthogonal to the remainder of the components, whereas in option 1 the eye component and frontal brain components are both estimated, not orthogonal, and removing of the eye component does not remove the frontal brain component.<br>
<br>
Option 1 is better, as it is less aggressive in removing brain components.<br>
<br>
If speed is a concern, you could do<br>
- ft_resampledata to e.g. 250 Hz or even less, estimate the components based on that and project them out of the original high Fsample data.<br>
- do ft_componentanalysis on a subset of the data (say every 4th data segment after cutting it in pieces), and project them out of the original segmented data<br>
- a combination of the two<br>
- try out anothe rica algorithm (fastica versus runica)<br>
- try out with the options of the ica algorithm, esp the stopping options<br>
- get a faster computer<br>
<br>
best regards<br>
Robert<br>
<br>
<br>
<br>
<br>
<br>
<br>
_______________________________________________<br>
fieldtrip mailing list<br>
<a href="mailto:fieldtrip@donders.ru.nl" target="_blank">fieldtrip@donders.ru.nl</a><br>
<a href="http://mailman.science.ru.nl/mailman/listinfo/fieldtrip" target="_blank">http://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br>
</blockquote></div><br></div></div></div></div>
<br>_______________________________________________<br>
fieldtrip mailing list<br>
<a href="mailto:fieldtrip@donders.ru.nl">fieldtrip@donders.ru.nl</a><br>
<a href="http://mailman.science.ru.nl/mailman/listinfo/fieldtrip" target="_blank">http://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br></blockquote></div><br><br clear="all"><div><br></div>-- <br><div dir="ltr">
Hamid R. Mohseni, PhD<br>Post-Doctoral Research Fellow<br>Institute of Biomedical Engineering<br>University of Oxford, OX3 7DQ, UK<br>Tel: +44 (0) 1865 2 83826<br></div>
</div>