Student Presenter(s): Mary Margarette Sanches, Dono Shodieva, and Angel Singh
Faculty Mentor: Niharika Nath
School/College: Arts and Sciences, Manhattan

Cervical cancer screening using pap smear staining and microscopy is a manual process that can be time-consuming and prone to human error. Our research investigates nuclear features to improve detection efficiency and classification. We examined the potential of indistinct nuclear shape (Nuclear Minor Axis and Major Axis Length) and texture features (Angular Second Moment (ASM), Contrast, Correlation, Inverse Difference Moment (IDM), and Entropy) to discriminate normal, moderate, and severe dysplastic cells of a benchmark dataset. We used Cell Profiler to extract the nucleus and its features, then conducted non-parametric tests, specifically Rank Sum and Kruskal-Wallis to determine statistical significance. Rank Sum results indicated that all texture features discriminated Normal vs Severe while most for Normal vs Moderate dysplastic cells (except for IDM).