Breast cancer histologic image cell segmentation

Hi everybody! I'm trying to detect both tumoral cells and linfocytes (darker and smaller points) in a dataset of images like the ones attached. I know that tumoral cells have a less omogeneous coloring and are bigger with respect to linfo. I need to segment these two structeres to obtain a mask that will be used for classification in the next step. I already tried a watershed algorithm and various combinations of thesholdings and morfological operations with a very bad final result (the algorithm doesn't detect a sufficient number of cells or detects agglomerations of them). I attach three images as an example of our data set. Thanks for all the replies!

1)

2)

3)

PS: these are some of our results (red=cancer, green=linfo)

1)

2)

3)

[Anand Karnam]

Hi, I am pursuing B.Tech in Instrumentation. If you don't mind, would you help me in my project, detection of breast cancer, I don't have the required dataset to work on.

[KALYAN ACHARJYA]

Please find the publicly avalible datasets

[Eleanor Betton]

Do you have an example of your code? Have you tried do any colour transformations to amplify the contast. Is detecting the cells or identifying between the tumoral cells and linfocytes which is the problem?

[KALYAN ACHARJYA]

Steps:

1. Do the Preprocessing (Contrast Enhancement)
2. Segmentation (Not so challenging as per attached images)
3. Compare the results with Ground Truth Images