Normalize by background

Hi, I need your help with an analysis

I need to measure intensity of cells stained in GFP, and compare the intensity between samples (or genotypes). The images are tiff, 8bit, RGB color, and green if I look at image mode when I open them with photoshop. I have also the corresponding DAPI images in blue. I generated a pipeline where I load the images, and I use the module ColorToGray for both images (GFP and DAPI), then I use IdentifyPrimaryObjects with the DAPI image in gray and then I use modules for MeasureObjectIntensity and MeasureObjectSize using all the time the gray images.
I know that my 2 genotypes have different levels of my protein stained in GFP, but when the pipeline shows me the original image in green and the change to gray, the genotype with lower levels of GFP shows what it looks like an autoscaling of intensity, because the background doesn’t look as negative as it looks in photoshop. My question is: is the pipeline doing an autoscaling? how can I compare between genotypes? If it is autoscaling, can I use rescaleIntensity dividing by image minimum? Will that be equivalent to substracting the background?

Thanks in advance


Hi Barbara,

When CellProfiler displays an image (either in a module display window, or by double-clicking on the image in the file panel), it shows it with the intensity normalized/autoscaled by default (i.e, lowest value mapped to 0, brightest value mapped to 1). If you want to see them in the original scaling (which I assume is what Photoshop is showing you), you can right-click on the image in CP and select “Image contrast > Raw” to display it in that format.

In any case, the intensity measurement modules always make measurements based on the raw intensity values, and any values exported will reflect this fact. So the important question is: do the actual intensity values reflect the expected differences in sample genotypes?