Le lundi 14 mai 2012 09:38:58 Paolo Zaffino a écrit :
> extract from it only the blood vessels
1. Your question is too vague.
2. You must first check whether the exam is of good quality, i.e, enough
contrast, acquired not too early, not too late, less than 1 mm spacing in 3
3. Use ROI to select ... a region of interest ... These will speed up
4. As a first approach, use Simple Region Growing to create a label map
(fiducial on contrast medium)
5. Use Model Maker to generate the 3D volume, then the Models volume to play
6. Use Reformat Widget tool to cross check the result. The result of Simple
Region Growing is dependant on a lot of factors, and is unpredictable. The 3D
model it generates may not be totally accurate. You should always cross check
the 3D model in the gray scale volume.
7. 4.1 is still crashing on a regular basis, 3.6.3 would be more incentive.
8. Don't expect what you can't do with Slicer, like a complete blood vessel
tree with the 'curve technique'.
9. It will take you months to get satisfied with what you do, still a cowboy ?
If you have a good quality AngioCT, your first good result will be very
pleasing, and hopefully, very helpful.
Le lundi 14 mai 2012 13:37:44 Daniel Haehn a écrit :
> the Slicer4 VMTK extensions
I found it to be great for normal blood vessels, and not at all for
pathological blood vessels.
Stenosis, calcified or not, are rendered in a way helpful for surgical decision
making (I know, Slicer always says it's not intended for clinical use. It's
just a legitimate disclaimer). At least I could not get satisfactory results
for the surgeon.