Optomised pipeline for images of different intensities

Hi all,

Currently I’m trying to measure the amount of nuclei( stained blue) in a population that are positive for BRDU (stained red). The problem I’m having is the intensities between each nuclei image. I’ve kept the exposure time the same when capturing them but some parts of the well have brighter background than others this is causing the pipeline I’ve created, to work within a narrow intensity band often rejecting images that are showing nuclei but are too weak in comparison to other brighter objects.

I’m hoping to find a way to normalise all the nuclei images background to each other, so that the brightest background images will be reduced and the pipeline will pick up all the nuclei in each image.
I’ve attached some examples of the difference between the backgrounds and the pipeline I’m using.

2brdu 17102013.cp (15.6 KB)
I’ve tried some rescaling and masking the image with some limited success.

Any help in improving the robustness of this pipeline would be greatly appreciated.



Hi guys,

Does anyone have any suggestions I could try to equalise the brightness of the background between the different nuclei images?

Anything I could try at this stage would be greatly appreciated.


As a possibility, I would remove the rescaling modules, and in turn suggest the following:

  • CorrectIlluminationCalculate: Set the input image as the raw input image, the smoothing filter as “median”, method to calculate smoothing filter size as “Manual” and the filter size as “200”. Leave the rest at default settings.
  • CorrectIlluminationApply: Divide the original image by the illumination function.

However, I strongly suggest working on your image acquisition protocol first, as your images are quite blurry and over-saturated. I don’t know how much control you have over this aspect, but any sort of image analysis quantification will be dependent on the input image quality, regardless of the robustness of the analysis itself.