| Re: XROI - how to invoke region growing [message #29232 is a reply to message #29230] |
Fri, 08 February 2002 13:44   |
Ted Cary
Messages: 53
Registered: October 2001
|
Member |
|
|
Hello Dr. Mueller,
I'll have to ask you to excuse my ignorance as well, but to help with this
I need to know exactly what you mean by "region growing." I don't have
experience with XROI or CT scans, but I did not find any mention of the
region growing function you are trying to invoke during a quick reading of
the XROI help files. If something like this exists, I could use it too.
I'm not sure if region growing has a special meaning in your field, but
here are my suggestions, for what they're worth:
If you want to simply scale the regions, you can call scale methods on them
individually, since they are just IDLgrROI objects. It looks to me as
though you can get them using the REGIONS_OUT keyword to XROI, scale them
all, then use the scaled ROIs as the REGIONS_IN in another XROI call if you
want. Of course if the regions are concave, then the margins of your new
"grown" regions might actually intersect the margins of the smaller regions
from which they were grown, which might not be what you want at all, since
the grown region won't contain all of the points in the smaller region.
If instead you want to "expand" or dilate the regions, maybe POLYFILLV the
ROI vertices and use DILATE. The array results will not be in a form you
can put back into an IDLgrROI, since there will be no margin vertices
identified nor a connectivity array, but they will allow you to extract
data from your scans easily, and you can continue to DILATE or ERODE the
regions at will. Unfortunately, unless you "hull" the interior points to
get the connectivity array and the margin vertices, you won't be able to
use XROI on the regions you've grown. This method will work fine if all
you want to do is 1) Draw ROIs in XROI, 2) "Grow" or "shrink" them, and 3)
Extract data from their interiors. You can even display the new margins
if you want by finding the contours of the interiors--you'll know every
point on the border in no particular order, but not every vertex. I do
this kind of thing all the time, and it should not require any or much
modification to those 50 pages of XROI code. Hope this helps.
"Dr. Stefan P. Mueller" wrote:
> Hi,
>
> I am new to IDL, so please excuse my ignorance:
>
> XROI looks like a good starting point for what I am planning to do:
> define a region-of-interest on a CT scan by region growing (to segment
> reproducible a lesion) and then go on to sovle my real problem.
> I just installed the demo for IDL 5.5 and ran XROI. I am able to
> display the CT Image. I can manually define all sorts of ROI's, and I
> can get statistics or a histogram. I have been unable, however, to
> figure out how to invoke the region growing function. Yes, I looked at
> the dokumentation too.
> As a beginner, trying to understand the code (probably over 50 pages)
> is daunting to me...
>
> Does anybody know what I have to do?
>
> Thanks a lot!
>
> Stefan
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
