I was able to use Filter - Other - Minimum with a particular set of parameters, followed by a Median filter to smooth the results, to get something not unlike what you showed, but given that I derived the parameters empirically from your very limited image, at exactly the size given, I wouldn't be surprised if the characteristics of this sequence of operations are way off from what you want.
In any case, if you'd like to see this "proof of concept" action to understand what I did, here it is:
Keep in mind this will be VERY sensitive to the thresholds and sizes chosen for the various steps, and is crafted to work at an image with just the resolution you showed above.
Thank you a lot for your proposition, It's not perfetc but it's a good (indirect) solution, and I can customize your action.
That was pretty clever Noel.
Glad I could help get you started, mdidish84. Are we diagnosing tumors or something here? Just curious.
Indeed I quantifies a subtype of cells in the pancreas, but the staining is very "light" on several image, and our usual methods to quantify do not work.
Your work helps me a lot for a first estimate !
Now I try to do a more detailled solution, with a script:
- for each pixel, I get color;
- If the pixel is black, I'm looking if there is a black pixel around (eg within 10 pixels)
- If this is the case, I draw a black line between the two black pixels
- from there, I can quantify the total area (lines inside) with an other software.
Just one problem, I know a little Java, but I never do script with photoshop; so I'm reading the documentation (Photoshop-CS6-Scripting-Guide).