Tutorial 57 - Nuclei (cell) segmentation in python using watershed

This video walks you through the process of nuclei (cell) counting and size distribution analysis in python. The process involves image segmentation using watershed followed by object measurements and saving results into a csv file for further analysis.

References:

[zafrirhavia]

Thanks sir
great video detailed and explained clearly.
I was looking for an explanation of watershקג and found your video and after that explanation I will go over your every video
thanks

[olivoto]

Congratulations on the amazing tutorial! I've been working on an example and have noticed that this works very well for round objects (works perfectly for coins for example). For thinner objects (like a knife), I always observe an oversegmentation (single objects are split into multiple ones). Any tip to work around this? Thanks!

[ibonholgado]

The 3D topographic image seems very interesting to explain many problems related to grey intensity.
Please, could you indicate which function you have used to obtain it, and if you have not used python, in which software is this function available?

[gustavofariasvelosocarvalh]

Hi friend, good evening. Everything is good?
I am following your tutorial and I am very grateful that you are sharing your knowledge with us.
I have a question, when I run the code I get the following error: Out[1]: <matplotlib.image.AxesImage at 0x7f2ce80a8fa0>
on the line: 36 ( plt.imshow(opening, cmap='gray') #This is our image to be segmented further using watershed
)
do you know how i can be solving this problem?
I use ubuntu on version 20 and so far, apparently I've installed all dependencies and all code so far has run without any problem.
Once again, thank you very much for your availability, kings regards.

[OhmFF]

Thank you, sir.
When I apply markers = cv2.watershed(image, markers), I got an error message error: (-215:Assertion failed) src.type() == CV_8UC3 && dst.type() == CV_32SC1 in function 'cv::watershed'
image is uint8 (1 Channel), and markers is also uint8. Would you mind telling me how to fix it?
Best regards,

[nayanjittalukdar]

where we get this image ?? can anyone tell me ?

[asmabenbrahem]

I have binary masks where there are some overlapped cells with different sizes, I tried many factors for the distance transform (a.k.a factor*max) but either I get the tiny cells and don't get the big ones separated or I get the big cells separated and the tiny cells disappears and don't detect them. Please help me sir if you have any idea about how to solve this problem.

[marwanabdou]

hi sir
when i aplied the distance transform i get this error message (error: OpenCV(4.0.1) error: (-215:Assertion failed) src.type() == CV_8UC1 in function 'cv::distanceTransform')
how can i fix it
thank you