An automatic system for cell nuclei pleomorphism segmentation in histopathological images of breast cancer

Nuclear pleomorphism is one of the criteria for diagnosing and grading breast cancer. The grading that is made by pathologist is subjective and prone to inter, intra observer variations. Furthermore, pathologists may need a huge time for evaluating all cases per day. Therefore, there is a necessity to provide an automatic system for a better diagnosis and detection. This paper proposes an automatic system for detecting and segmenting cancerous nuclei, which is partly different from healthy nuclei segmentation systems. In contrast, our system detects critical nuclei with any shape, border and chromatin density even in higher scores. This system avoids segmenting healthy cell nuclei. It only detects and segments a high percentage of deformed cell nuclei, which are necessary for nuclear pleomorphism scoring even cells with vesicular nuclei that are not detected in any other algorithms.