Request help for brightfield segmentation

It is very difficult to select the patches on the image, which is attached.

Thanks a lot.

@sqzhang

Just a couple things:

  1. You should not save your data as .jpeg files. Always save in the proprietary format of your imaging system or as .tiff files. You can read more on that here.

  2. I would point you to the plugin: Trainable Weka Segmentation (TWS). It comes with Fiji. NOTE: Fiji is Just ImageJ - it is simply a distribution of ImageJ that comes with a bunch of plugins bundled - ready for you to use out-of-the-box. If you are just getting started, we recommend downloading/using Fiji. I ran a quick test on a cropped section of your image using the TWS default settings:

Hope this helps!

@etadobson

Thank you so much. I have tried Weka, but could not find the appropriate training features for this kind of images, because the background intensity is not homogeneous. Could you please tell me what features you have applied?

@sqzhang

Sorry for the delay… was away on holiday. You can correct the background (see links below) in this case - will improve training, etc.


https://imagej.nih.gov/ij/plugins/background.html

Also search here on the forum for background correction threads!

Thank you so much for your kind help. @etadobson

1 Like

Hello,
Please excuse the interruption but if you make a copy of your image then Edit>Invert the copy and use Process > Image Calculator and subtract the Image from the Inverted Image you will effectively double the dynamic range of the Image, which may help.
Bob

I got curious and tried the attached image using a similar method (pixel classification), which is in our SW called Intellesis (oviously a commercial solution with trial). And for you whole image it worked quite well despite the uneven illumination.

image

and the analysis output:

image

Sebi (from ZEISS)

1 Like

You have uneven illumination from the top left to the bottom right of the image. I ran a line profile with a line width of 50 pixels from the top left to the bottom right and got this line

profile. This will interfere with your backgound subtraction and segmentation. See the information on flat field correction here.

I should have noted that I had to kludge together a macro to do this. One has to set the line width manually and then run the macro. There should be a better way to do this… Anyway, this wasn’t pretty but it worked…

plot_profile.txt:
open("/path-to-image/image.jpeg");
lw = 50;
w = 1380;
h = 920;
setLineWidth(lw);
makeLine(lw, lw, w-lw, h-lw);
run(“Plot Profile”);