Hello
Can I select the pixel resolution of my segmentation?
is it the same as native image resolution?
thanks
Binary labelmap representation has the same resolution as the master volume that you select first.
Before you start segmenting your volume, use “Crop volume” module to crop it to the minimum necessary size, enable isotropic voxel size (check Isotropic spacing), and reduce voxel size (e.g., set Spacing scale to 0.5). Use this cropped and resampled volume as master volume for segmentation.
2 Likes
I see a cropping ROI but it seems that you haven’t actually cropped the volume. Was that intentional? If you use a smaller ROI then you can have much finer resolution volume (so you can segment small, thin structures very nicely) without having a large volume. Also note that you can smooth the segment (using Smoothing effect) and/or you can apply smoothing during labelmap->3D surface conversion (using Show 3D button’s Set surface smoothing option).
1 Like
an off topic question: if I want to use this selection, BUT INVERTED, to create a mask in all volume (then to mask all that is outside of this selection)… can i dublicate the selection? and invert it?
You can use Logical operators effect to copy and invert segments.
yes thanks, worked perfectly
for some structures is easiest to select the structure, then invert and mask all outside
I am working with a scene with several volumes with different slice thicknesses. I’m trying to understand the resulting number of pixels in the segmentation/segments saved as nrrd.
Above it is stated that “Binary labelmap representation has the same resolution as the master volume that you select first.” Is that in the Segmentation Module?
Assuming it is the Segmentation Module, shouldn’t the scene hierarchy determine the initial selection? It seems random to me, sometimes set by scene hierarchy and sometimes set by last selection.
I would like to have everything in one scene and have the segmentation number of pixels match my volume. Any tips with multiple volumes?
One problem that I encountered is that the minimum pixel value (by segment statistics) is often lower than my threshold value. I assume due to interpolation. How is thresholding performed when volume number of pixels < segment number of pixels?
Thanks in advance.
In Segment Editor module.
It is the source volume that is chosen first in “Segment Editor” module. There is no scene hierarchy there. The choice is arbitrary, but it is not random, but this is what most users would expect. You can change the geometry anytime by clicking on the “Specify geometry” button.
You can move around segments with any geometry between segmentations. They will be automatically resampled as late as possible (when the segmentation is modified or saved to disk). You can force resampling or changing resolution using “Specify geometry”.
The source volume is always resampled to the current segmentation’s geometry.