Enhancing Breast Cancer Diagnosis Using Explainable Models

Abstract

Breast cancer is a primary cause of mortality among women across the world, and early diagnosis and proper classification of the subtypes are of vital importance in enhancing the survival rate. The paper is a review synthesis of recent research developments in explainable artificial intelligence (XAI) with machine learning (ML) and deep learning (DL) models to diagnose, predict, and segment breast cancer based on 20 peer-reviewed articles published between 2023 and 2025. The significant themes are the use of XAI methods, including SHAP and LIME and Grad-CAM, to make medical imaging processes such as mammography, ultrasound, MRI, and histopathology images more understandable, and genomic and clinical data. The analyzed literature shows a higher level of diagnostic accuracy (between 84 and 99.5) and attempts to overcome such issues as data imbalance, high dimensionality, and clinical trust by visualizing features of importance and using attention mechanisms. This review shows the transition to clear AI systems that are in line with clinical processes, where there are gaps in multimodal integration and real-world testing.