Applying 3D ROI to a non-segmented / thresholded image?

[Snider, Kaitlin]

Hi :)


I'm fairly new to ImageJ, but I've spent a lot of time learning how to use the various available plugins etc over the last several months.  I have confocal .lsm Z-stacks in three channels, one of which is a nuclear stain.  I've worked with the Trainable Weka Segmentation to automate segmentation of nuclei vs. background, and then used binary, watershed, and analyze particles to get ROIs for each nuclei in each slice.  With the 3D ROI manager (http://imagejdocu.tudor.lu/doku.php?id=tutorial:working:tutorial_for_3d_roi_manager) I can take the segmented image and get great 3D ROIs.  What I'd like to do from here is use those 3D ROIs to then measure intensity values in the other two channels - un-segmented and un-thresholded.  The 3D ROI manager specifies that images should be segmented before loading, and trying it with the raw .lsm file just freezes the program (same if I convert to a raw .tiff file).  I can save and re-load the ROIs in the 3D ROI manager, but trying to load the 3D ROIs in Fiji's standard ROI manager onto the raw .lsm file just gets me an error.  Any ideas?  Is there a button I'm missing, or another plug-in that would help transfer the 3D ROIs to the raw image?


Link to the raw .lsm image: https://drive.google.com/file/d/0Bw4MsPRmUoyISWU2aGNRalBRNjQ/edit?usp=sharing

Link to the segmented binary nuclei image: https://drive.google.com/file/d/0Bw4MsPRmUoyIUnFPWkNheXV4cVk/edit?usp=sharing

Link to the 3D Objects file for the segmented image: https://drive.google.com/file/d/0Bw4MsPRmUoyIVjFuS0VNNjA1NTg/edit?usp=sharing

?Thanks very much for your time :)


Kaitlin Snider

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Thomas Boudier]

Hi Kaitlin,

The 3DManager requires either a binarized image or a segmented one. A
segmented image is actually a image where each object has a distinct pixel
value (see count mask in analyze particles). This segmented image can be
obtained from the binarized image either with 3DObjectCounter or "3D
segmentation" in 3DManager (with threshold 255-255).
I did "3D segmentation" on your segmented binary image, then "Add Image"
using the segmented image and it´s working fine, you have 2078 objects,
unfortunately even with a 2D watershed, they are not well separated in 3D.
The "quantif 3D" button will use the active window to measure signal
intensities within selected objects (if no objects are selected), all
objects are used.

Do not hesitate if you have more questions.

Best,

Thomas

> Hi :)
>
>
> I'm fairly new to ImageJ, but I've spent a lot of time learning how to use
> the various available plugins etc over the last several months.  I have
> confocal .lsm Z-stacks in three channels, one of which is a nuclear stain.
>  I've worked with the Trainable Weka Segmentation to automate segmentation
> of nuclei vs. background, and then used binary, watershed, and analyze
> particles to get ROIs for each nuclei in each slice.  With the 3D ROI
> manager
> (http://imagejdocu.tudor.lu/doku.php?id=tutorial:working:tutorial_for_3d_roi_manager)
> I can take the segmented image and get great 3D ROIs.  What I'd like to do
> from here is use those 3D ROIs to then measure intensity values in the
> other two channels - un-segmented and un-thresholded.  The 3D ROI manager
> specifies that images should be segmented before loading, and trying it
> with the raw .lsm file just freezes the program (same if I convert to a
> raw .tiff file).  I can save and re-load the ROIs in the 3D ROI manager,
> but trying to load the 3D ROIs in Fiji's standard ROI manager onto the raw
> .lsm file just gets me an error.  Any ideas?  Is there a button I'm
> missing, or another plug-in that would help transfer the 3D ROIs to the
> raw image?
>
>
> Link to the raw .lsm image:
> https://drive.google.com/file/d/0Bw4MsPRmUoyISWU2aGNRalBRNjQ/edit?usp=sharing
>
> Link to the segmented binary nuclei image:
> https://drive.google.com/file/d/0Bw4MsPRmUoyIUnFPWkNheXV4cVk/edit?usp=sharing
>
> Link to the 3D Objects file for the segmented image:
> https://drive.google.com/file/d/0Bw4MsPRmUoyIVjFuS0VNNjA1NTg/edit?usp=sharing
>
> ?Thanks very much for your time :)
>
>
> Kaitlin Snider
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
>

--
Thomas Boudier, Associate Prof. UPMC,
IPAL UMI 2955, Singapore.

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html