can not recognize the nuclei of image

[jacobchenzx]

hi, Volker, thanks for your work! I have some issues when using the Intensity Ratio Nuclei Cytoplasm Tool. following your guide, I loaded the two images of 8 bit, and then set the names, as you could see.


but when I run 'S' button, it can not recognize the nuclei of the image like your guide, and the result of intensity, area of cytoplasm and nuclei do not make sense. Do you know how I could resolve it? thanks a lot.

[volker-baecker]

Hi @jacobchenzx,

you only need to open one of the channels, the other one will be opened automatically. I see that you entered "FLAG" as name of the cytoplasm channel, but the name on the window actually is "FALG". Please correct and try again. Should it still not work, please send me a pair of original input images.

Best,
Volker

[jacobchenzx]

hi，@volker-baecker
thanks for replying! you're right that the name 'FLAG' does not match. I found it later but it does not work well, either.
when I tried again, I found the original input image is 8-bit with tif format!!! After transforming into jpg format, it worked well. So the problem may be the format of the input images. I wonder which format of picture could work well.
here are the results, does it work well, right?

the result shows the whole intensity of nuclei and cytoplasm. if we want the result of the intensity of each cell, there is other tool named the cellprofiler.

[volker-baecker]

Hi @jacobchenzx,

the file format should not be the problem, however the two images must be in the same folder and have exactly the same name except for the two parts that you enter in the options and that are used to find the corresponding images.I would not recommend to convert to jpg. Tif is actually the native format for ImageJ. If you send me a pair of tif files, I can check. I tried with my own tif files and it worked fine.

The individual cells are often hard to segment and separate from each other. Cellprofiler has this very interesting Propagation method , that combines gradients with the distance map ((Jones et al, 2005). We also use cellpose to get the individual cells.

[jacobchenzx]

Hi @volker-baecker
if the file format is not the problem, it makes me confused. Here are my input data in tif format. oops, GitHub does not support uploading the tif file. could I get your email address?
that' true! the individual cells in some pictures are hard to segment when using the same paraments. so somebody also uses the intensity of perinuclei to replace the cytoplasm. thanks for recommending the cellpose, I'll try it.

[volker-baecker]

Hi @jacobchenzx,
your image has a pixel size of 0.0138889 inch set in the meta-data. That makes that a nucleus has an area of about 4 square-inches. The default min. area in the tool is 20. That's why no nuclei are found. When you convert to jpg, the metadata about the pixel size is lost and the area is measured in pixel. You can achieve the same effect without converting, by selecting remove-scale in the options of the tool.
Best,
Volker

[jacobchenzx]

‘You can achieve the same effect without converting, by selecting remove-scale in the options of the tool.’ it works!!!! I appreciate that you are so kind to answer my question and upload so many useful tools!!!!