Comparison of Histomorphology of Variants of Epithelial Malignancies of the Uterine Cervix Using Computer Aided Image Analysis.

Abstract

Tissue diagnosis of cervical cancer in small biopsies can be challenging. Our goal is to assess the differences in the 
characteristics of the cells of the various types of epithelial malignancies of the uterine cervix using computer aided 
image analytic tool. We did an observational study of uterine cervical biopsies diagnosed as epithelial malignancy. 
The cases were diagnosed in the department of Morbid Anatomy and Histopathology of UniOsun Teaching 
Hospital, Osogbo from January 2006 to December 2015. The biopsies went through routine tissue processing in the 
laboratory. The slides were scanned using a CS2 APERIO digital slide scanner using standard conditions. The cell 
detection tool of the QuPath software was optimized for cell detection and measurements. The various attributes of 
the nucleus and cytoplasm were then measured. The nuclear measurements show that the size of the nucleus of cells 
of mucinous carcinoma and adenocarcinoma is larger than that of other cells. The mean nuclear circularity is higher 
in cells of adenosquamous carcinoma and poorly differentiated squamous carcinoma compared to other variants. 
The mean nuclear eccentricity of cells of well differentiated and moderately differentiated squamous cell 
carcinoma are higher than that of the other variants. The nuclear hematoxylin optic density standard deviation is 
highest in mucinous carcinoma. The cells of squamous cell carcinoma have higher average eosin optic density. 
Computer-aided image analysis using QuPath software can be useful in distinguishing the various types of 
epithelial malignancies of the uterine cervix.