It seems to me you do not necessarily need ilastik for this task.
Although, if you insist on using it. I believe the Object classification workflow is what you need.
You can import your nuclear masks in the
segmentation image section at the first step. Of course, your raw data of the “2nd channel” in
Raw data section.
speckles are not segmented. You will have to train your classifier in the following step. Once done, you may only need to export the object feature table to
.csv or whatever you want.
I don’t really understand what you want to do with the 2nd channel. Just counting the number of speckles ? If so,
number of holes seems to be close to what you need.
To note, I’m not sure about the last part, you may have to invert the image before counting.