A Novel Framework for Automated Cervical Cancer Detection and Prediction using Graph Convolutional Neural Network-based Models
Subramanian Pitchiah Maniraj, Eswararao Boddepalli, A. Anitha, RVS Praveen, Seeniappan Kaliappan, P. Chandra Sekhar Reddy · 2025
Cervical cancer, the biggest cause of cancer death, affects 570,000 women worldwide. The HPV virus causes most abnormal cervix growth. HPV screening and testing have reduced mortality rates in developed nations, while developing nations continue struggle with low healthcare costs and low oncologist-patient ratios. The suggested computer-aided cervical cancer screening technique includes improvement, extraction, selection, and classification. Enhancement decreases noise and sharpens contrast to highlight cervical anomalies. Identifying normal and abnormal cervical cells by form, texture, and intensity is feature extraction. Identifying the most important qualities improves classification accuracy and reduces computing complexity with feature selection. In the last phase, ML classifies cervical cell samples as carcinogenic or not. This systematic technique enhances early detection accuracy in low-resource healthcare settings and is feasible, affordable, and effective. The suggested technique classifies cervical cancer better using machine learning and feature selection. This diagnostic instrument is efficient and affordable for underprivileged locations. Early identification boosts cervical cancer survival. The evaluated system lowers diagnostic costs and improves access in underdeveloped nations. Future research should focus on AI-driven automation to improve clinical accuracy and practicality.