<html>
<head>
</head>
<body bgcolor="#FFFFFF" text="#000000">
Thanks Mark + Jörn for the replies!<br>
I'll have a look at the NAI. <br>
<br>
On 27.03.2012 15:48, "Jörn M. Horschig" wrote:
<blockquote
cite="mid:203022dc538e40168d5b72ded3195dcc@EXCCAHT-1.mail.uke.ads"
type="cite">
<meta http-equiv="Content-Type" content="text/html;
charset=ISO-8859-1">
<meta name="Generator" content="MS Exchange Server version
08.03.0213.000">
<title>Re: [FieldTrip] one-sample cluster test</title>
<!-- Converted from text/plain format -->
<p><font size="2">Dear Ulrich,<br>
<br>
<br>
First of all, you might want to use the Neural Activity Index
(NAI), as<br>
also described in the tutorial:<br>
<a moz-do-not-send="true"
href="http://fieldtrip.fcdonders.nl/tutorial/beamformer#source_analysiswithout_contrasting_condition">http://fieldtrip.fcdonders.nl/tutorial/beamformer#source_analysiswithout_contrasting_condition</a><br>
<br>
Then, to answer your question, I am tempted to say that this
would be an<br>
invalid approach when dealing with power, so in
frequency-space. Power<br>
obviously has a lower bound, so that the average of any random<br>
collection of power values will never be 0. </font></p>
</blockquote>
<br>
That sounds plausible for raw power data, but actually, I want to
compare whether my power change to baseline activity is different
from zero (maybe that wasn't evident in my first post). I think
that's a correct approach.<br>
<br>
<br>
<br>
<blockquote
cite="mid:203022dc538e40168d5b72ded3195dcc@EXCCAHT-1.mail.uke.ads"
type="cite">
<p><font size="2">Though, what you are<br>
suggesting is the same as a regular t-test, isn't it?<br>
In case you are dealing e.g. with am LCMV beamformer, I am not
quite<br>
sure, but the null-hypothesis you suggested (amplitude==0)
sounds fair.<br>
<br>
In any case, building a surrogate distribution same mean and
variance<br>
than your data might be a better way to deal with your
problem. That<br>
should be easily doable in Matlab without any need to
implement this in<br>
FieldTrip. And note that the MC problem is always present, but
a<br>
cluster-based correction (similar to Bonferroni, but
correcting by the<br>
number of voxels in the cluster not by the number of all
voxels) sounds<br>
legitimate to me. Afair, that's how it's done in FT.<br>
</font></p>
</blockquote>
Are you sure? I am not aware of any correction to the alpha level
that is done by the cluster-based correction in FT. Afaik all it
does is look for a certain number of significant voxels that are
close in time, freq and space.<br>
Best, Ulrich<br>
<br>
</body>
</html>