Yeast segmentation in brigh tfield images


I want to segment yeast cells in bright field images. I attached the pipeline I am using at the moment. It works with binary images and the results are good for easy images where the cells are isolated. Unfortunately the results get worse when the images get more complicated for example when there are lots of cells in one image or when the cell shape starts to change in budding processes.

Do you have any idea how to improve the segmentation in these images? Maybe working with grey scale images? Or would edge detection make any difference?


Brightfield_analysis0910.cp (8.89 KB)

Hi ML,

Yes, grayscale images are much more likely to give you better results. I loaded up your pipeline, but hesitate to even look past the first few modules because it looks much harder to declump than I suspect it would be if you started with grayscale images. Can you post a grayscale one?


Hi David,

thanks for your reply.
I uploaded the bright field version of the image and a more easier image for testing.

Do you have any suggestions on how to do a proper preprocessing to identify a big amount of cells? Then the odds are higher to receive a high number of segmented cells after filtering…