Automatic Cell Counting and Percentage of Live and Dead

[krura sandhu]

Hello,

I am trying to make an automatic cell counter macro that can be extended to give results for percentage of live vs dead cells. So far it has worked well but I have 2 problems with my macro.

The first problem encountered is that for some images, the cells are on a grid and the macro auto local thresholds the grid with the cells. This causes the macro to analyze and count the grid components as cells. I have thought of using remove outliers and it mostly gets rid of the grid but also removes vital cells for cell counting. Additionally, even after using the remove outliers there are still some grid components left. Any suggestions to make the grid disappear before the macro auto local thresholds it will be great. I have thought of using Gaussian Blur but I do not know if that will be effective. Below is the image of the cells on the grid and the auto local threshold image after using remove outliers.





The 2nd problem encountered is that the macro cannot count live cells and dead cells. So far I have allowed the multichannel image to split into 3 channels and discarded the blue. From there I would do a "Process Find Maxima" for each channel which counts the live cells and dead cells. However for some images, the Noise Tolerance has to be set high. This causes some cells to not have any points located on them.  Is there a way to allow ImageJ to automatically count live cells and dead cells so this problem can be avoided?


Solutions and suggestions to any of these problems will be great. I am very new to ImageJ so I apologize in advance for the lack of technical language.

From Krura