Deep Learning and Ensemble Approach for Cervical Cancer Detection using Medical Image Analysis

Abstract

Cervical cancer remains a significant global health challenge and is one of the leading causes of cancerrelated mortality among women, particularly in developing regions where access to early screening and diagnostic facilities is limited. Early detection plays a crucial role in improving survival rates, with over 90% success in early-stage diagnosis. However, conventional histopathological examination is timeconsuming, labour-intensive, and subject to inter-observer variability, leading to inconsistent results. To address these challenges, this study proposes a Deep Ensemble Learning (DEL) approach named Cervical Cancer Detection Network (CVD-Net). Initially, histopathology images undergo preprocessing techniques including resizing, normalization, contrast enhancement, and noise reduction to improve image quality. Traditional machine learning models such as Logistic Regression (LR), Gaussian Naive Bayes (GNB), and Quadratic Discriminant Analysis (QDA) are used as baseline classifiers. The proposed method leverages Convolutional Neural Network (CNN)-based feature extraction combined with a Random Forest Classifier (RFC) to automatically learn high-level spatial and textural features, minimizing reliance on handcrafted features. The model classifies images into four categories: Squamous Cell Carcinoma (SCC), Negative for Intraepithelial Malignancy (NILM), Low-grade Squamous Intraepithelial Lesion (LSIL), and High-grade Squamous Intraepithelial Lesion (HSIL). Experimental results demonstrate superior performance of the proposed approach, achieving an accuracy of 97.5%, precision of 97.9%, recall of 96.6%, and F1-score of 97.1%. These findings highlight the effectiveness of the proposed system for accurate and early detection of cervical cancer, potentially aiding clinical decision-making and reducing diagnostic errors.