how to get infected regions count

I have grayscale image of red blood cells in which some of the cells are infected by malaria that means some portion of it is darkened, how should i get the count of infected and non-infected cells?

Answer 1

[Image Analyst]

Usually when people ask for image processing advice, they insert an image into their post with the frame icon.

http://www.mathworks.com/matlabcentral/answers/6200-tutorial-how-to-ask-a-question-on-answers-and-get-a-fast-answer

So all I can guess at is this:

1. Find and count all the cells by thresholding with a high threshold.
2. Then set a different, lower threshold to find the dark regions.
3. If some cells have more than one dark region, merge all dark regions within a cell.
4. Then call bwlabel to count the blobs, both normal and infected.

[Ravi Dange]

Sir, This is the image which I am trying to work on, This is showing RBCs and some of those are infected by malaria i just want to get the count of infected cells i have used regionprops to get count of all cells but lagging in evaluating no. of infected cells. Please suggest solution.

[Image Analyst]

Do color segmentation to find the purple cells. Fill them in. There are some color segmentation demos in my File Exchange. http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862&sort=downloads_desc

Then split apart touching and overlapping cells following Steve's blog: http://blogs.mathworks.com/steve/2013/11/19/watershed-transform-question-from-tech-support/

Now do color segmentation for the dark purple things. Mask it by the light purple cell mask.

Next use imreconstruct to use the dark purple mask to pull out only those light purple blobs that have dark purple in them.

Call bwlabel() on that and the cell mask to get counts of both the number of light purple cells and light purple cells that have dark purple stuff in them.

Easy. Post your code if you get stuck.

[Ravi Dange]

thanks for the answer... I'll try this way...