An intelligent approach for the automated segmentation and quantification of Immunohistologically stained nuclei

Abstract

The manual monitoring process of cancer cells development is a subjective, time consuming process, as it typically relies on the visual recognition and experience level of the pathologists. An automated nuclei segmentation and quantification in Immunohistologically stained images have remained a challenging task. Previous methods used have shown an oversight in the segmentation and counting of the two different types of stained nuclei within the same image (i.e. positive brown stained nuclei and negative blue stained nuclei). Our exclusive method addresses this issue by producing an automated means for the segmentation and counting of nuclei based on the monochromatic characteristics of the different types of stained nuclei objects. Ultimately this could aid pathologists towards more accurate and time efficient diagnosis by considering the affects of protein antibodies inside the nuclei. Our experimental work has proven to produce promising results. This was through the appropriate allocation, segmentation and counting of nuclear contents inside Colonic Cancer of Immunohistologically stained images.