Diffusion-Enhanced Magnified Histopathological Images for Robust Breast Cancer Classification
Yan Chengxiao, Xiaoyang Zeng, Awais Ahmed, Muhammad Hanif Tunio, Fan Shuhuan · 2023
Histopathological Image Analysis (HIA) is essential for breast cancer diagnosis, where classifier accuracy and precision are crucial for patient care. While developing a breast cancer classifier, researchers faced data limitations and the need for superior histopathological image discrimination. Complex cancerous tissues require nuanced and detailed analysis, but diverse and representative datasets are scarce. This study proposes a naive solution using diffusion techniques to propose a robust cancer classification method (Diffusion-enhanced BCCNet) with an efficient image enhancement mechanism. The three public Breast Cancer datasets (BreakHist, BHI Breast Histopathology Images, and BACH) have shown significant progress, demonstrating the potential of our proposed methodology to address data availability issues.