<div dir="ltr"><div>Thank you very much, Andras.<br><br></div><div>Jiahui<br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Mon, Oct 6, 2014 at 11:44 AM, Andras Lasso <span dir="ltr"><<a href="mailto:lasso@queensu.ca" target="_blank">lasso@queensu.ca</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">





<div link="#0563C1" vlink="#954F72" lang="EN-US">
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">See this long thread about this issue:<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><a href="http://slicer-devel.65872.n3.nabble.com/BRAINS-rigid-registration-td4032229.html" target="_blank">http://slicer-devel.65872.n3.nabble.com/BRAINS-rigid-registration-td4032229.html</a><u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">Short summary: There were some performance regressions in mutual information metric computation, threading, etc. that have been recently fixed, so use the latest
 ITK version. ITKv4 uses a slightly slower but more accurate optimizer, but if you adjust the number of samples and stopping condition then usually the registration time only increases by about 30-50% (and you might get better quality results).<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">You can see all the changes that have been done in 3D Slicer and BRAINS repositories to go from a 10x-100x slowdown to an approximately equivalent registration
 speed with ITKv4 compared to ITKv3.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">Andras<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><b><span style="font-size:11.0pt;font-family:"Calibri","sans-serif"">From:</span></b><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""> Community [mailto:<a href="mailto:community-bounces@itk.org" target="_blank">community-bounces@itk.org</a>]
<b>On Behalf Of </b>Jiahui Wang<br>
<b>Sent:</b> Monday, October 06, 2014 10:27 AM<br>
<b>To:</b> <a href="mailto:insight-users@itk.org" target="_blank">insight-users@itk.org</a><br>
<b>Subject:</b> [ITK] [ITK-users] Speed of ITKv4 based Gradient Descent Optimizer<u></u><u></u></span></p><div><div class="h5">
<p class="MsoNormal"><u></u> <u></u></p>
<div>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt">Dear All,<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal">I modified ImageRegistration8.cxx to run my own rigid registration.<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal"><br>
However, I found that the processing time of ITKv4 based RegularStepGradientDescentOptimizer is almost 10 times more than that of ITKv3 based one.
<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt"><br>
The program was run on the same machine. The only difference is the ITK libraries. 
<br>
<br>
Could anyone give me some suggestions to solve this issue?<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt">Thanks,<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal">JW<u></u><u></u></p>
</div>
<div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
</div>
</div></div></div>
</div>

</blockquote></div><br></div>