How to use Fiji to automatically analyze GFP expression levels for each cell?

Hello, all! I have taken a few pictures of my GFP expressing cell line with Leica inverted widefield microscope. I’m trying to analyze the intensity of each cell’s GFP expression level. However, when I try to use the particle analysis tool, the program will recognize a lot of noise as cells. The reason for that, I think, is the GFP expression levels varies greatly between cells, so by setting a threshold would cause imageJ to include the background noise. I could just ignore the really bright cells or ignore the dim cells in the threshold setting, but I really want to use an automated process instead of boxing out cells myself. Is there another tool in Fiji that could achieve this? I will include a picture I took. Thank you so much!
The image in google drive:
https://drive.google.com/a/uw.edu/file/d/0Bzg3HJnv9iTPZWJURlQ1dXNVaWc/view?usp=sharing

1 Like

Hi @BenZ -

So… just taking a quick look at your images… your background levels are very high. Not only that - but it varies across the image. (also - worried that this difference in background across your image may also influence your GFP signal of the cells themselves??) Take a look at this snapshot I made of a line profile using Plot Profile across your image… a good thing to play with to see differences between what you consider signal and background. From this I think you can see why you are having troubles picking up some cells versus background - across the image - some are have the same signal.

Is there any way that you can improve the signal-to-noise ration during acquisition? This will make your life a lot easier in the end when processing your datasets.

Others will have more technical insight to provide you I’m sure… But hope this at least gives you a start on planning for this analysis.

eta :slight_smile:

1 Like

Too @BenZ

It’s always better to use a different label to first mask your cells using a membrane/cytoplasmic marker or general cell stain/dye that doesn’t impact your GFP expression. Then you will have homogenous staining of those cells - and in theory - a strong signal-to-noise ratio. You use this to create a mask of your cells … and then can measure the GFP signal within.

That would be the best practice. It is never a good idea to create a mask on the signal you are expecting to change depending on expression … Read more on that topic here from another forum thread.

eta :slight_smile:

1 Like

Hello Eta, thank you so much for the replies! I’ll look into your suggestion on masking the cells.

Best,
Ben

no worries @BenZ … and if you have more questions - just post here. We are here to help make sure you get some sound data. :slight_smile: