<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
</head>
<body bgcolor="#ffffff" text="#000000">
<font face="Helvetica, Arial, sans-serif">Hi,<br>
I've got a question concerning the registration framework. More
precisely, it's an issue concerning the metric I'm using to perform a
registration of two images.<br>
<br>
For my application, I've taken and slightly modified the
DeformableRegistration15 example one can find among ITK's Registration
examples.<br>
<br>
The metric used is the MattesMutualInformationImageToImageMetric. My
questions is related to the SetNumberOfSpatialSamples method of the
metric. According to the ITK Software Guide, a low number of spatial
samples (relative to the number of pixels in the image) yields poor
results, which I could verify.<br>
<br>
However, I was stunned when I found that omitting the call to
SetNumberOfSpatialSamples yields results comparable to the ones
provided by a large number of spatial samples (at this point I did not
pay attention to the computational cost).<br>
This behavior seems strange to me.<br>
<br>
Previous to calling the aforementioned method, the ReinitializeSeed
method is called, but as far as I've seen, it's got little implications
on the above behavior.<br>
<br>
I'm wondering about the implications of invoking the </font><font
face="Helvetica, Arial, sans-serif">SetNumberOfSpatialSamples method </font><font
face="Helvetica, Arial, sans-serif">(or skipping its call).<br>
<br>
For sure there must be something that I did not understand.<br>
<br>
Could someone please cast some light on this issue?<br>
<br>
The images I'm using are CT images, with a resolution of 512x512 and a
number of slices of around 100.<br>
I'm using ITK 3.18<br>
<br>
Thank you.<br>
<br>
Regards,<br>
JON HAITZ</font>
<div class="moz-signature"><br>
</div>
</body>
</html>