HybridBoost Ensemble Model for Enhanced Cervical Cancer Detection: A Multi-Stage Feature Optimization Approach

Early cervical cancer diagnosis drastically improves treatment efficacy, requiring sensitive and effective models for diagnostics. This research offers a new HybridBoost Ensemble Model that integrates cutting-edge preprocessing techniques, feature extraction, feature selection, and classification to boost detection efficiency. The proposed preprocessing pipeline contains TriPhase Augmentation with normalization, grayscale trans- formation, and flipping, in combination with Pentamorphic Transformation through brightness, contrast enhancement, edge detection, Gaussian blur, and rotation to optimize image quality. This unified histogram of oriented gradients (HOG), local binary pattern (LBP), Scale-Invariant Feature Transform (SIFT), and Gabor filters method, enhanced gradient-texture fusion (EGTF), helps to represent feature exhaustively. To reduce the features with dimension reduction and preserve vital information, this uses a Genetic Algorithm (GA) and PCA. The HybridBoost Ensemble Model is a union of logistic regression, random forest, support vector machines, KNN, Adaboost, and GradientBoost classifier, optimized through GridSearchCV. The ensemble ap- proach augments bagging, boosting, stacking, and voting to improve generalization and minimize overfitting. Experimental evaluations exhibit substantial performance improvements over baseline methods measured by various performance metrics.. The proposed approach is an effective, automated, and scalable solution for early cervical cancer detection.