Marker-controlled watershed segmentation of nuclei in H&E stained breast cancer biopsy images

In this paper we present an unsupervised automatic method for segmentation of nuclei in H&E stained breast cancer biopsy images. Colour deconvolution and morphological operations are used to preprocess the images in order to remove irrelevant structures. Candidate nuclei locations, obtained with the fast radial symmetry transform, act as markers for a marker-controlled watershed segmentation. Watershed regions that are unlikely to represent nuclei are removed in the postprocessing stage. The proposed algorithm is evaluated on a number of images that are representative of the diversity in pathology in this type of tissue. The method shows good performance in terms of the number of segmented nuclei and segmentation accuracy.