How can i get the Eccentrecity, Primeter, Convex area, Area, and Solidity by using imfindcircles

Please, i have problem, which is founding Eccentrecity , Primeter , Convex area , Area , and Solidity from using of imfindcircles for all detected blood cells in the attached colored microscopic image.

or how can i using the results from imfindcircles to compute any thing by regionprops function

Answer 1

[Image Analyst]

You can't get that from imfindcircle() - it's not meant for that. You will need to identify the cell areas and then use regionprops(). There's no way around it.

[Hany Elsalamony]

 Thank you very for your great response, but how can i transform the resulted cells only into binary image *bw* to applying regionprops for all their areas, ..., etc

rgb = imread('D:\cell shapes and images\sicklecellanemia.jpg'); gray_image = rgb2gray(rgb); bwkj=im2bw(rgb); imshow(gray_image)

[centers, radii, metric] = imfindcircles(rgb,[15 18], ... 'ObjectPolarity','dark','Sensitivity',0.90,'Method','twostage','EdgeThreshold',0.2); figure,imshow(rgb) [centersBright, radiiBright, metricBright] = imfindcircles(rgb,[15 18], ... 'ObjectPolarity','bright','Sensitivity',0.97,'EdgeThreshold',0.2);

benigncells=length(centers)+length(centersBright)

hBright = viscircles(centersBright, radiiBright,'EdgeColor','Gr');

h = viscircles(centers,radii,'EdgeColor','Gr');

[Image Analyst]

You need to segment out the cells. I'd probably start by extracting the blue channel, thresholding, and throwing out blobs smaller than a certain size (because they must be noise, not cells). You can size filter either by using bwareaopen() or regionprops(). See my Image Segmentation Tutorial for an example of that. http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862

[Hany Elsalamony]

Thanks alot, yes i know this; but if there exist any way to working with the cells, which are already extracted by imfindcircles , please tell me. because i built all my code but this point only stopped me, which is getting the data for all extracted cells like areas,...etc. thanks again for your efforts and response

[Image Analyst]

No, there is not. If you simply converted the circles to a binary image, all your eccentricities and solidities would be exactly the same - probably not what you want or else why measure them? You expect them to be different, and they are on the actual shape of the cells, bot no ton circles.

[Hany Elsalamony]

In fact, i want to apply some classifiers on these exported data. Now,in the same way, if i have different radii and centres can i draw circles or disks in another image to applying regionprops?How?.

Thank you all the time