Identifying the Particles
In the first three episodes of this tutorial, the groundwork has been laid to carry out the real task at hand, namely to identify the spots in this image as viral particles and then to count them. In this post, I’ll describe the process I went through to configure each of the many settings in this module. The bottom line is that the process is an iterative trial and error.
Here’s a screenshot of the IdentifyPrimaryObjects module settings that I finally ended up with.
The input image is the output image from the previous module that I named “CorrGreen”. This module doesn’t produce an image, but a set of objects. In my case, I called them ‘Virus’. After trying a few different thresholding strategies, I had an idea the spots in this image that were virus had a diameter of between 1 and 12, but I didn’t want to throw away particles outside this range. Additionally, I did want to discard objects that were touching the edge of the image. The online help is excellent at describing the different thresholding strategies. “Automatic” worked well in this situation as a thresholding strategy. I used intensity instead of shape to differentiate touching objects and intensity as the method to draw lines between touching objects. I left all the radio buttons ticked as “yes” except the retain outlines of objects, and I let Cellprofiler fill holes after thresholding and declumping. Again, I tried to change the threshold strategy and the declumping strategy, but found that the ones chosen aobve gave the best results for this image. See below for the output from this module that lets you instantaneously decide if your parameters are optimal.
The output looks like it could be giving a good result, but I need to have a closer look. The magnifying glass allows me to zoom in to any particular part of the image to see how the thresholding performed. I’ve done just that in the second image below. After careful consideration, I feel that the parameters for this module are optimal.