Several questions about ExampleSparklePipeline


I have several questions about ExampleSparklePipeline, because my trial also need counting.

Fisrt, why are there two images 1-162hrh2ax2.tif and 1-162hrhoe2.tif in this pipeline. Do they have relation to each other? In my every trial, I have only one image to run.

Second, in the second IdentifyPrimAutomatic modual, what do the different colors stand for in the identified h2ax image?

Finally, in the Relate module, what is the difference between the original sub object and new sub objects. In original sub object, what does the difference in color imply? Why produce an new sub object?

In my trial, I need counting sparkles, so I feel the ExampleSparklePipeline may be fit for me. According to my sample image displayed below, how can I coordinate the settings? Need I add or delete the modules?

Truly, thanks a lot!



I’ll answer your questions in order:

(1) The reason we have two images is that 1-162hrh2ax2 has the speckles highlighted while 1-162hrhoe2 has the nuclei which contain the speckles highlighted. By having two images, it becomes much easier to determine which nucleus the speckles come from.

(2) The colors in the second IdentifyPrimAuto module serve to illustrate the identified objects; it helps to keep in mind that the color is not a measured quantity but just distinguishes one object from another. For the upper right panel, each identified speckle gets a different color so you can tell them apart.

(3) The Relate object matches “parent” objects (in this case, the hoe2 nuclear images) to the “children” objects (the h2ax speckles). Each child object is spatially contained with its parent object. What makes the new sub-objects different from the original sub-objects is that now these relationships have been identified in the new ones. The child speckles are now colored so that those that have the same parent have the same color, and you can see which speckles belong to which nucleus.

Without two images, the problem faced in the image you uploaded is how to identify the bright green spots (speckles) from the dim green spots (nuclei?). If you had two channels as in the example pipeline, it becomes easier. With the image you have, it’s bit more difficult.

Looking at your image, my suggestion is using ColorToGray to split the image into the three color channels. I think you’ll see that the speckles seem to be bright in the red and blue color channels, whereas the larger objects (nuclei?) are visible only in the green channel (you should check this yourself, since the superimposed colored arrows may be misleading me). At this point, you now have two images to work with and you can use the same approach as in the example pipeline.

If you’re just interested in counting speckles, then the Relate module isn’t necessary and you can just run IdentifyPrimAuto on the color channel image which has just the speckles and use ExportToExcel to export the results. But if you want to determine which speckle belongs to which object, the Relate module will be useful for you.

Hopefully, this will get you started.


Thanks, your remark is helpful for me.


Another suggestion - your ability to identify the dots within the cells in your images might be improved by using the SmoothOrEnhance module with the “Enhance BrightRoundSpeckles” option. Use the image that is produced by that module for the IdentifyPrimAutomatic module.


I can’t find the SmoothOrEnhance unit in image processing model from CellProfiler.exe I download from your website.
Why? I downloaded it in March. I don’t konw whether there is a SmoothOrEnhance unit in old version.
Need I download a newer version? If do, whether will the old version be deleted?



The latest compiled version is 1.0.5122 and was released in March. You might have just missed the release date and picked up an older version. If so, go ahead and download the latest version, replacing the older files.

If you’re not sure what version you have, go to Help -> Getting Started and see what number is listed.