Count cells and messure the area

[Jule_B (Jule)]

Hey,

I am trying to count cells and messure the area of the cells and clusters with Fiji. I have a stack of pictures and they all look similar to this one:

I already solved the light gradient problem with the Bandpass Filter and now I am stucked trying thresholding. I tryed a couple different filters and thresholding methods. The best result I got was with substract Background and then using otsu threshold. Then it looked like this:

Does anyone knows a better method, where I would not loose the structure and the Area of the cells?

Thanks a lot.
Jule

[Research_Associate (MicroscopyRA)]

You may want to try CellPose on the images that have been Bandpass filtered. There was a post here showing brightfield results.

Does anyone have an importing masks script corresponding to the working export tiles script

@melvingelbard I just wanted to say thank you for this, as it has made importing CellPose results very clean. I need to maybe work on some overlap issues, but I can now tile and save, process in CellPose, which saves the results as png files in the same folder, then cycle through the png results in the folder to create objects, and voila, brightfield segmentation that is not completely terrible! [image] And in QuPath, I can color code them by circularity. Also thanks to @petebankhead for the…

From the paper

MouseLand/cellpose

Note that you can run it on the website to test it before going through the trouble of setting up a Python environment for batch processing (which you still may want to do if you have hundreds of images).

If you want to put the labeled image back into FIJI, @waisman has created a plugin for that as well.

Development of LabelsToROIs FIJI/ImageJ plugin

Development of LabelsToROIs plugin There have been many recent machine learning tools created for image segmentation and object detection that work astoundingly well, such as the amazing #cellpose . In many cases, these algorithms generate their output in the form of labeled images. For a computer scientist or a user well versed in image analysis using programming languages, the use of label images is common practice, and thus they can rapidly incorporate these algorithms into their everyday rou…

[Dobrokhotov1989 (Oleg)]

Some time ago, I used PHANTAST plugin for ImageJ to separate cells from the background: (GitHub - nicjac/PHANTAST-FIJI: Repository for the FIJI/ImageJ implementation of the phase contrast microscopy segmentation toolbox (PHANTAST)). After getting a mask with PHANTAST you can multiply it by your original image. Thus you’ll preserve the structural features of cells and cell clusters, while the background will be 0.