Robust Training of Deep Learning Models for Mammogram Classification

Josué Martínez-Martínez, Olivia Brown, Jun Bai, Sheida Nabavi

Abstract: Deep neural networks are increasingly applied in healthcare to detect and diagnose medical conditions, including breast cancer through mammogram analysis. However, these models are vulnerable to adversarial attacks and distribution shifts, undermining diagnostic reliability and trust. This study explores baseline methods to address these vulnerabilities and proposes a modified robust training approach to enhance model robustness and reliability in mammogram analysis.

External IDs:dblp:conf/isbi/Martinez-Martinez25