Posted by
Thomas Boudier on
Jul 30, 2014; 11:13am
URL: http://imagej.273.s1.nabble.com/3DObject-Counter-Surface-Map-tp5008891p5008988.html
Hi Nadia,
The labelling in 3D-Manager uses a different algorithm, that basically
stores every object in 3D, and 3D-OC segment in 2D for all slices and then
reconstruct the 3D. However the results should be exactly the same. Maybe
the algorithm used by 3D-Manager requires more memory. Once the segmented
image added into 3DManager, the list of 3D objects is displayed with
obj(i)-val(j), j is the grey-level of the object in the segmented image.
Clicking on a object in the list will display its contour in the slice of
the current stack (maybe you need to click LiveRoi first). Hope this
helps.
Best,
Thomas
> Hi Thomas,
> thanks for your help, that is very useful.
> I think I will use the 3D Manager to do this though, I find it is much
faster. Are there big differences in how these two plug-ins segment? Im
wondering because computation time is so different. Also, after
segmenting in 3D Manager (not using 3D-OC), how can I match the value
assigned to each object to objects on the map (eg to delete certain
points)?
>
> Thanks again,
> Nadia
> ________________________________________
> From: ImageJ Interest Group [
[hidden email]] on behalf of Thomas
Boudier [
[hidden email]]
> Sent: Friday, July 25, 2014 3:02 PM
> To:
[hidden email]
> Subject: Re: 3DObject Counter Surface Map
>
> Hi Nadia,
>
> 3D object counter should display both volume and surface from the
detected objects, do not forget to adjust brightness in order to see all
labelled objects. The surface should represent the boundary of your
objects, you can then binarize this image (threshold) and then add it or
use color merge to see the contours superimposed on your original
images. You may also want to have a look to 3DManage that does the job
for you :
map at the end looks nothing like what I would expect looking at the
thresholded image. Instead of displaying the rendered surfaces (which I
need so I can compare the analysis/segmentation with my original image)
it
>> just has tiny dots, similar to what I would expect for the "centroids" to
>> look like.
>>
>> Have I misunderstood the descriptions of the various maps generated?
How can I be confident that the analysis has picked up the mitochondria
correctly if I cant compare the final segmented "image" with the
original?
--
Thomas Boudier, Associate Prof. UPMC,
IPAL UMI 2955, Singapore.
--
Thomas Boudier, Associate Prof. UPMC,
IPAL UMI 2955, Singapore.
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html