Measure localization of spots

I am new to using cell profiler. I have practiced the basics of it.
But i need help and some insights with my task.
I have 3 channels of images.
1- nuclei

2- signal

3- cytoplasm

My task is to measure the localization of spots(signal) around the nuclei in the cell.
Can i have some idea about solving this task.
I know that i have to identify the primary, secondary and teritary object and relate them somehow. But i am not clear about it. Need help.
I am also attaching the images i have.
working on the task i have come up with the attached pipeline.
Can i have the pipeline checked and know if i am heading the right direction. pl1.cpproj (219 KB)
I am also attaching the output of relate objects here.


Hi,

I would suggest taking a look at the speckle counting pipeline on our examples page. The basics of that pipeline seem to come close to what you want to do.

Hope this helps!
-Mark

Thank You for the guidance.
Using it as reference i put up the attached pipeline. relatedots.cpproj (328 KB)
at the end i have the dots related to cytoplasm.



i have to make some measurements
1)closest distance of the dots to membrane
2)distance of the dots to cell centroid
3) distance to nuclear centroid
Can you guide me.

Regards
yuva

[quote=“yuvashree”]
i have to make some measurements
1)closest distance of the dots to membrane
2)distance of the dots to cell centroid
3) distance to nuclear centroid[/quote]

Using the RelateObjects module, you can do all three. However, you’ll have to use the RelateObjects module twice, once for the spots as children and nuclei as parents and select “Centroid” for “Calculate child-parent distances” (for #3), and the other for spots as children and the cell as parents and select “Both” for “Calculate child-parent distances” (for #1 and #2). However, #3 will only work if the nuclei overlap the spots; if they don’t, the distances will not be counted.

Then in ExportToSpreadsheet, the distances will be contained in the per-spot spreadsheet.

Regards,
-Mark

Thanks for the timely help.
I need some help with the measurements.
I dont know how to get this done.

‘‘Radius in pixels to include 5%, 10%, 15%, 25%,50%,75% of all remaining spots in a cell’’
’‘Number of spots within 20, 40, 80. 120 pixels from a given spot centre’’

Regards
Yuva

No module can give you precisely this measurement. However, the MeasureRadialDistribution module can give you something like it. You would need to add the following modules:

  • ExpandOrShrinkObjects to shrink the spots to a point.
  • ConvertObjectsToImage to convert the shrunken spots to a binary image.
  • MeasureRadialDistribution with the binary image as the input image, and the cells as the input objects. Specify the number of bins as the number of concentric rings to divide the cell into.

What MeasureRadialDistribution does is give you fractional intensities from an image, computed across a series of rings starting from the cell center, proceeding radially out to the cell boundary; see the module help for more details.

One of the measurement made is the fraction of the total intensity at a given radius (FracAtD), for each radius. Since the input image is a binary image (i.e, 0 or 1) of the spots, the intensity calculated is actually proportional to the spot count. So you can cumulatively sum up the radial measurements for each object to get the cumulative fraction of the total number of spots for each cell.

The MeasureObjectNeighbors module provides the number of neighbors for each object (in this case, the dots). Select “Dots” for both the objects to measure and the neighboring objects, select “Within a specified distance” for the method, and enter the number of pixels desired. Since you have multiple distances to compute, you will need a separate MeasureObjectNeighbors module for each one. The number of neighbors for each dot will be written as part of the per-dot spreadsheet.

Regards,
-Mark

Thank You.
Just a quick question. Is it possible to do this in matlab already.

Regards
Yuva

Hi Yuva,

We don’t have any Matlab code to do this. But feel free to write your own Matlab code! :smile:

Cheers,
David

On this same context how do i mesaure ‘‘Distance to cell outline projecting nuclear centroid’’

Thank You

Regards
Yuva

Hi Yuva,
I’m not sure what you mean by this measurement.
-Mark

Hi Yuva,

Maybe this measurement in MeasureObjectSizeShape is what you need?

David

Hi guys.
Thanks for the great software, we’re truly happy to have the possibility to get NUMBERS out of image data for reviewers.

Well, we have one article pending where a reviewer asked us to support our statement about perinuclear localization of vesicles in transgenic cell line vs. controls. Knowing each cell is a bit different in size simple nucleus-vesicle distance would maybe not be sufficient. Rather, I’ve figured out that if one would express the localization of a vesicle as a distance of nucleus-vesicle vs distance of membranme on that same vector one could remove absolute dimensions and summarize huge amount of cells.

We label our cells with DAPI and segment the cells using either a plasma membrane lable or phaco images, and have thousands of images available. Provided we would know how to make this work the problem would pretty much be solved=article revised.

Looking forward to interesting solutions!

Greetz,

m

Hi meeksNTG,

This sounds exactly what the module MeasureObjectRadialDistribution was designed for. The idea is to divide an object (say, cytoplasm) into annuli/rings extending out to the object boundary using either the centroid of the larger object or another primary object (e.g. nucleus). You can choose to scale the rings to the size of the cell which it sounds like you do want (so select “Yes” when asked to “Scale the bins?”). If you don’t want the intensity of the channel to play a role, you would have to threshold/binarize the object prior to this module, but you certainly don’t have to. If you only want the count of objects falling within the ring, you could shrink each previously segmented vesicle down to a point using ExpandOrShrinkObjects and then the MeasureObjectRadialDistribution “intensity” will be equivalent to the object count.

Does that help?
David

Hi again,

I want know if exist a example pipeline with this process ?

Thanks a lot,

Isabela.

Hi @Isabela_Possebom,

Our example pipelines are available at https://cellprofiler.org/examples. If that doesn’t help, could you describe what you’re looking for a bit more?

Best,
Pearl