<div dir="ltr"><span style="font-family:arial,sans-serif;font-size:14px">Dear list,</span><div><font face="arial, sans-serif"><span style="font-size:14px"><br></span></font></div><div><font face="arial, sans-serif"><span style="font-size:14px">I got a response from Tim about this issue. Let me forward it to share with you.</span></font></div>
<div><font face="arial, sans-serif"><span style="font-size:14px"><br></span></font></div><div><div><font face="arial, sans-serif"><span style="font-size:14px">%%%%%%%%%%%%%%%%%%%%%%%</span></font></div><div><span style="font-family:arial,sans-serif;font-size:14px">Piecewise linear detrending should not really be regarded as a "filter" per se. It is intended to remove trend and ultra LFOs within a sliding window that is approximately the same as that which will be used for subsequent VAR modeling. </span><div style="font-family:arial,sans-serif;font-size:14px">
<br>A principal advantage of using a detrending operation rather than a HP filter is that there is no filter delay or phase shift, and no temporal smoothing or ringing. The disadvantage is that it is less straightforward to precisely control the frequency response of the transformation (i.e. determine which frequencies are suppressed vs. passed). Additionally, one must consider the overlap between windows (which are added, with a sigmoidal weighting function to smoothly transition between windows).<br>
<div><br></div><div>When replacing a filter with detrending, one must bear in mind that the frequency "cutoff" depends on the sliding window size and overlap properties. In particular, I would make the window size sufficiently large that it spans no less than a 1/2 cycle of the lowest frequency of interest. </div>
<div><br></div><div>Another option when combining with SIFT is to simply ignore the detrending option in pre-processing and select linear detrending in the modeling step. This will apply a linear detrender to each windowed time-series before fitting the VAR model. This is a principled step to remove the 1st order (mean) local non-stationarity.</div>
<div><br></div><div>I might suggest that some further empirical investigation into the properties of the piecewise detrending operation vs. filtering would be a worthwhile endeavour for someone to take up.</div><div><br>
</div>
<div>Tim</div><div>%%%%%%%%%%%%%%%%%%</div><div><br></div><div>Thanks again Tim for explanation.</div><div><br></div><div>Makoto</div></div></div></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">2014-03-24 17:42 GMT-07:00 Makoto Miyakoshi <span dir="ltr"><<a href="mailto:mmiyakoshi@ucsd.edu" target="_blank">mmiyakoshi@ucsd.edu</a>></span>:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Dear Mengyan,<div><br></div><div>One of the problems of a filter is that it causes 'ringing'.</div>
<div><a href="http://en.wikipedia.org/wiki/Ringing_artifacts" target="_blank">http://en.wikipedia.org/wiki/Ringing_artifacts</a> </div>
<div><br></div><div>If you ask me whether the piecewise detrending can substitute high-pass filter-I don't know. That seems a good question. Let's ask Tim Mullen.</div><div><br></div><div>Makoto </div></div><div class="gmail_extra">
<br><br><div class="gmail_quote">2014-03-22 7:25 GMT-07:00 诸梦妍 <span dir="ltr"><<a href="mailto:bj12116@gmail.com" target="_blank">bj12116@gmail.com</a>></span>:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div><div class="h5">
<div dir="ltr">Dear eeglablist,<div><br><div>I‘m confused about the relationship between DC-detrend and high-pass filter.</div><div><br clear="all"><div>According to </div><div><a href="http://sccn.ucsd.edu/wiki/Chapter_6.5._Preprocessing" target="_blank">http://sccn.ucsd.edu/wiki/Chapter_6.5._Preprocessing</a></div>
<div>DC-detrend is an alternative to high-pass filter, as they both could remove drifts. </div><div><br></div><div>Is this mean I needn't use DC-detrend after a 0.1Hz high-pass filter?</div><div><br></div><div>My recordings were amplified through BrainAmp DC amplifiers with a bandpass filter of 0.016–100 Hz, digitized on-line with a sampling frequency of 1000 Hz. Then the data was processed and segmented to -200 to 800ms epochs.<br>
</div><div><br></div><div>Any suggestions are welcome.</div><div>Thank you in advance!</div><div><br></div><div>Mengyan Zhu</div><span><font color="#888888"><div><br></div><div><br></div><div><br></div>-- <br>
Mengyan Zhu<br>Psychology department, Peking University<br>
Dormitory 2061, Building 48,No.5 Yiheyuan Road, Haidian District, Beijing 100871, China <br>
E-mail: <a href="mailto:bj12116@gmail.com" target="_blank">bj12116@gmail.com</a>
</font></span></div></div></div>
<br></div></div>_______________________________________________<br>
Eeglablist page: <a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html" target="_blank">http://sccn.ucsd.edu/eeglab/eeglabmail.html</a><br>
To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu" target="_blank">eeglablist-unsubscribe@sccn.ucsd.edu</a><br>
For digest mode, send an email with the subject "set digest mime" to <a href="mailto:eeglablist-request@sccn.ucsd.edu" target="_blank">eeglablist-request@sccn.ucsd.edu</a><span class="HOEnZb"><font color="#888888"><br>
</font></span></blockquote></div><span class="HOEnZb"><font color="#888888"><br><br clear="all"><div><br></div>
-- <br><div dir="ltr">Makoto Miyakoshi<br>Swartz Center for Computational Neuroscience<br>Institute for Neural Computation, University of California San Diego<br></div>
</font></span></div>
</blockquote></div><br><br clear="all"><div><br></div>-- <br><div dir="ltr">Makoto Miyakoshi<br>Swartz Center for Computational Neuroscience<br>Institute for Neural Computation, University of California San Diego<br></div>
</div>