Detection of Nuclei Cell in Histopathological Images of Uterine Cancer: Adenocarcinoma of Endometrium

Abstract

Pathologist visual inspection of Histopathological uterine tissue sample is still considered as confirmatory test for uterine cancer finding. Manual assessment depends upon the knowledge and experience of pathologist. Hence subject of concern is objective analysis. An effective way of diagnosis, grading and classification of endometrium adenocarcinoma is structural analysis of the tissue sample. The effectiveness of computer assisted diagnosis, grading and classification depends on glandular structure and detection of individual nuclei cell. Hematoxylin channel from H&E stained image is extracted using color deconvolution algorithm. Morphological operation and thresholding are carried out in this work for structural analysis of glandular region and for detection of nuclei cell. To rise the effectiveness of the system elimination of cell nuclei from stroma is done before feature extraction is also described.