XGBoost-Based Breast Cancer Diagnosis System for Multiclass Mammographic Density and Mass Region Detection

Abstract

Introduction: Early detection is crucial in improving survival rates for women as breast cancer is a main cause of death. This study suggests an XGBoost-Based Breast Cancer Diagnostic System created for multiclassifying mammographic density and detecting mass regions. The system was tested on two popular mammographic datasets, INbreast and DDSM, and was pitted against various models like Convolutional Neural Networks (CNNs), Random Forest (RF), Support Vector Machines (SVMs), Logistic Regression (LR), and K-Nearest Neighbors (KNN). XGBoost consistently performed better than these models on all important measures, such as accuracy, precision, recall, F1-score, and AUC, delivering the top performance on both datasets. Its capacity to grasp intricate feature relationships and efficiently manage multiclass issues makes it ideal for this diagnostic endeavor. The suggested system offers a precise, effective, and strong tool for automating breast cancer diagnosis, with substantial potential for improving clinical decision-making.