Counting Cells automatically in ImageJ

3 posts / 0 new
Last post
bigwoof
bigwoof's picture
Counting Cells automatically in ImageJ

Hi There,

I am working with algae and need to perform routine cell counts to determine my algal culture concentration.  I can use the cell counter plugin to count the cells by hand but I have not found a way to do this automatically.  Is there a plugin or tutorial to do this.  I presume that it has been done before. 

If you want an example of an image there is one here:
http://dl.dropbox.com/u/419501/Count%201.JPG

Cheers

Phil

20101975
20101975's picture
U need to check it on various

U need to check it on various databases of research articles if this has been done or not. In ImageJ there is a facility of counting such entities, provided you have to kark them and the software will count automatically accordin to the marker selected.
Same thingcan also be done with StereoInvestigator.

The FFM
The FFM's picture
The link to the file in the

The link to the file in the drop box is now dead.

However if you can take a fluorescent image of your cells so that each has a uniform out line then you will be able to perform the following procedure in ImageJ

put 10ul of cell in a haemocytometer and take the image using a UV microscope equipped with the appropriate excitation/emission filters.

1. Open image file with ImageJ software.
2. Convert to binary image (Process > Binary > Threshold)
3. Count cells (Analyze > Analyzed Particles... check Display results, Clear results table, Summarize).
4. Repeat counts with additional large squares to be sure that the cells are evenly distributed and to minimize stochastic error.  (The standard deviation is ~ sqrt(mean)

you obviously need to have your image file calibrated so that you know the area of the region that you are counting to calculate cells/ml

For a brightfield image like this one

1. Open image file with ImageJ software.
2. Select Image > Adjust > Color Threshold 
click the "Threshold" check box and adjust the sliders on the saturation histogram to see just the outline of the cells
3. Analyze>Analyze particles

set minimum pixel size to 10 or more pixels and experiment with the circularity settings and select how you want the results output in the results window and on the image

below i did a very rough count that isn't perfect.  With the parameter optimization you will be able to get a true count.