<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
          "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html  xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en"><head>
<title></title>
<meta http-equiv="content-type" content="text/html;charset=utf-8"/>
<meta http-equiv="Content-Style-Type" content="text/css"/>
</head>
<body>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt">Dear Robert / other fieldtrippers,</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt">I’m looking for advice on the following.</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt"><br />
</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt">I’m trying to do clusterrandanalysis on TFR data with the cfg.
statistic=’actvsblT’ option. The tutorial ‘Cluster-level randomization tests’
gives an example in which separate TFRs are calculated for activation and
baseline data. The clusterrandanalysis.m documentation also states that two
datasets must be passed. </span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt"><br />
</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt">However, I’d like to make the activation vs baseline comparison work for
baseline and activation TF windows identified in single TFRs that include a
prestimulus and a poststimulus period. To do so, I used cfg. latency, cfg.
frequency and cfg. keepindividual with freqgrandaverage to create separate
datastructures for the baseline and activation timefrequency windows of
interest.     </span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt"><br />
</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt">Unfortunately, this doesn’t work. I get an error message:</span></font></div>
<div align="left"><font face="Arial" color="#228b22" size="2"><span style="font-size:10pt">??? Error using ==> clusterrandanalysis</span></font></div>
<div align="left"><font face="Arial" color="#228b22" size="2"><span style="font-size:10pt">Not all time points in the first data set are in the activation (post-stimulus)
period.</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt">This is probably due to clusterrandanalysis expecting a positive time axis for
the activation data and a negative time axis for the baseline data. Hence,
although I defined TF windows of equal length for activation and baseline, it
looks like Fieldtrip is unable to ignore the original, but irrelevant, time axis
definition. </span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt"><br />
</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt">I considered reading in the same data twice (through freqgrandaverage) with
different definitions of the time axes. However, this would still require
clusterrandanalysis to deal with activation and baseline TF windows that are
not matched in time on the time axes of their respective datastructures. So I
suspect it would still not work.</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt"><br />
</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt">Any advice on how I could work around this would be greatly appreciated.</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt"><br />
</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt">Best wishes,</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt">Peter Praamstra</span></font></div>
<div align="left"><font face="Arial" size="2"><span style="font-size:10pt"><br />
</span></font></div>
<div align="left"></div>
</body>
</html>