For the hippocampus dataset, we directly obtained the segmentation results by the Watershed algorithm from the original publication [20]. For the simulation and NSCLC datasets, we implemented the Watershed algorithm using the nucleus images based on scikit-image [38] and SciPy [39] python packages. Specifically, we segment each image by (1) performing morphology erosion on the image to enable separations of cell instances that are adhesive together, (2) computing the distance map of the image and generating a set of markers as local maxima of the distance to the background, and (3) feeding the image and markers into the watershed algorithm in scikit-image for segmentation.
Do you have any questions about this protocol?
Post your question to gather feedback from the community. We will also invite the authors of this article to respond.