CERVIA: A Hybrid Deep Learning Framework With Explainable  AI for Automated Cervical Cancer Classification From Pap Smear Imagesys

CERVIA is a novel hybrid deep learning framework designed for automated classification of cervical cancer from Pap smear images. Accurate detection of cervical abnormalities is critical for early diagnosis and effective treatment, yet traditional manual examination of Pap smear slides is labor-intensive and prone to human error. To address these challenges, CERVIA integrates three complementary deep learning models: HybridDenseNetInception, which captures multi-scale hierarchical features; KaryoFormer, a transformer-based model that leverages global context and attention mechanisms; and CytomeshModel, which models the structural relationships between cellular components. This hybrid architecture enables the framework to exploit the unique strengths of each model, achieving superior performance in identifying normal, precancerous, and malignant cells. In addition, CERVIA incorporates explainable AI techniques that highlight the morphological and cytological features influencing classification, thereby improving interpretability and clinical trust in automated predictions. Using the CytologCervicalCancer dataset from Kaggle, the system demonstrates high accuracy, sensitivity, and robustness, proving its effectiveness as a reliable tool for early cervical cancer screening. The framework not only reduces the diagnostic workload for pathologists but also provides transparent, interpretable results, facilitating better clinical decision-making. Overall, CERVIA represents a significant step forward in AI-assisted cytopathology by combining high performance with explainable insights.

Keywords: Cervical Cancer, Pap Smear Images, Hybrid Deep Learning, Explainable AI, HybridDenseNetInception, KaryoFormer, CytomeshModel, CytologCervicalCancer, Automated Screening, Medical Image Classification.