Unfortunately, it's necessary to then edit slices individually, move seeds etc.
So the point set and point set interactor are all useful.
I might have a "play" next week.
M
On 18 May 2012, at 10:02, Kilgus, Thomas wrote:
Hi Matt,
I think I got your point there. But if you want to render so many points per
slice, wouldn’t it be better to render something like a “seed image” (e.g. a
binary mask image)? It’s the same with segmentations. You could mark every
pixel with a point, but for simplicity you would use a binary image. Or is it
necessary to interact with each seed point?
Regards
Thomas
From: Clarkson, Matt [mailto:[email protected]]
Sent: Freitag, 18. Mai 2012 09:57
To: Kilgus, Thomas
Cc: mitk-users
Subject: Re: Point Set Rendering in 2D
Hi there,
I understand your point about better usability, and the fact that it does help
you find points quicker in 2D. But when there are lots of points, it looks very
confusing.
My use case is that we have an algorithm that goes roughly as follows:
1. User places seeds on a slice, and does a 2D region growing for that slice.
2. User hits a button "Propagate Up"
3. Project the current set of seeds on the current slice to the Next slice
Perform 3D region growing for the region Forward of the current slice
with that set of seeds, and the current thresholds.
4. For each slice Forward of the current slice, extract the 2D slice
5. Perform connected component analysis to extract each individual component.
6. Place a seed at the centre of mass of each connected component.
Thus, depending on thresholds you may end up with MANY points per slice.
I am happy to make any immediate changes, to provide the desired behaviour, as
I need it. I wanted to know how MITK people would prefer it to be done.
At the moment I cannot offer to refactor it to VTK.
Thanks
Matt
On 18 May 2012, at 08:41, Kilgus, Thomas wrote:
Hi Matt,
I believe that this is meant for better usability. The crosses are rendered
smaller, to illustrate that a point is coming on a slice near to the current
one. Thus, the points are easier to locate and there should not be much
confusion about them, since they look different compared to “real” points.
Can you explain your use case? When is it mandatory for you to render the
points only and their exact location and no helper objects around to locate
them?
We have a plan to re-write this mapper and port all GL code to VTK. I guess
this would change the look a little and could even result in your desired
behavior. But I think we would like to keep this effect for usability purposes
and, in addition, we currently have no resources for this project.
Regards
Thomas
From: Clarkson, Matt
[mailto:[email protected]]<mailto:[mailto:[email protected]]>
Sent: Donnerstag, 17. Mai 2012 18:35
To: mitk-users
Subject: [mitk-users] Point Set Rendering in 2D
Hi there,
when points are rendered as crosses in 2D, they are also shown on previous and
next slices, but a bit smaller.
I believe that in mitkPointSetGLMapper2D.cpp this is caused by:
Vector3D diff=p-projected_p;
ScalarType scalardiff = diff.GetSquaredNorm();
and
if((m_ShowPoints) && (scalardiff<4.0))
{
I would like to have crosses only visible on the slice to which they correspond.
How would people suggest modifying this? An extra property to control whether
we draw them on prior/next slices? Or something else?
Thanks
Matt
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
mitk-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mitk-users
