Decoding Breast Cancer Detection: An Explainable AI Framework for Transparent Diagnosis

Breast cancer remains a leading cause of mortality among women worldwide, necessitating early and accurate detection for improved prognosis. This study presents an Explainable AI (XAI) framework for breast cancer detection, leveraging advanced machine learning (ML) algorithms to classify tumors as malignant or benign based on digitized histopathological features. We employ Support Vector Machines (SVM), Random Forests, K-Nearest Neighbors (KNN), and Decision Trees, optimizing hyperparameters using Grid Search Cross-Validation to maximize predictive performance. Among the tested models, SVM with a radial basis function (RBF) kernel achieved the highest accuracy of 98.24%, with a recall score of 96.77%, minimizing false negatives which is a critical factor in cancer diagnosis. To enhance model interpretability, we integrate SHAP (Shapley Additive Explanations) and LIME (Local Interpretable Model-agnostic Explanations), identifying key predictors such as perimeter, compactness, symmetry, and smoothness of cell nuclei. A DNN is further explored to improve performance, achieving a recall of 93.75%. The results underscore the potential of XAI-driven ML models in clinical decision-making, offering both high diagnostic accuracy and transparency. Our approach bridges the gap between black-box AI models and medical interpretability, paving the way for trustworthy, explainable AI-assisted cancer diagnostics.