Dhidik Prastiyanto, Elma Herdiyanti Putri, Ahmad Fashiha Hastawan
Cervical cancer is among the most prevalent malignancies affecting women. This condition can be identified through a Pap smear test. Pap smear image data is utilized to detect abnormalities in these cells. A reliable segmentation algorithm is required to identify the contours of the nucleus and cytoplasm accurately. The experimental process involves three key stages: preprocessing, initial Segmentation, and contour segmentation. During preprocessing, the image regions are preliminarily grouped into nucleus, cytoplasm, and background using the Fuzzy C-Means algorithm. Once the initial contours are extracted, Segmentation is further refined using the RGGVFS method. Radiating Generalized Gradient Vector Flow Snake (RGGVFS) enhances the existing Gradient Vector Flow Snake (GVFS) by introducing an external force algorithm based on the Radiating Edge Map (REM). Experimental results demonstrate that the proposed method provides more precise nucleus detection compared to prior approaches. The average accuracy and Zijdenbos Similarity Index (ZSI) for nucleus segmentation reach 0.993 and 0.902, respectively. Meanwhile, for cytoplasm segmentation, the average accuracy and ZSI are 0.896 and 0.884, respectively. © 2025 Author(s).
Faculty of Engineering, Universitas Negeri Semarang, Semarang, Indonesia
