Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Breast Cancer Classification with Enhanced Interpretability: DALAResNet50 and DT Grad-CAM

Suxing Liu

TL;DR

This work tackles the challenge of accurate and interpretable breast cancer classification from histopathology images under data imbalance. It introduces DALAResNet50, a ResNet50-based classifier augmented with a lightweight dual-activated attention module, and DT Grad-CAM, a visualization method with adaptive thresholding. Across BreakHis, BACH, and Mini-DDSM, DALAResNet50 achieves superior accuracy, F1, IBA, and GMean compared with eight baselines, while DT Grad-CAM provides clearer, more localized heatmaps. Together, these contributions enhance diagnostic performance and transparency, supporting clinical decision-making and broader medical imaging applications.

Abstract

Automatic classification of breast cancer in histopathology images is crucial for accurate diagnosis and effective treatment planning. Recently, classification methods based on the ResNet architecture have gained prominence due to their ability to improve accuracy significantly. This is achieved by employing skip connections to mitigate vanishing gradient issues, enabling the integration of low-level and high-level feature information. However, the conventional ResNet architecture faces challenges such as data imbalance and limited interpretability, which necessitate cross-domain knowledge and collaboration among medical experts. To address these challenges, this study proposes a novel method for breast cancer classification: the Dual-Activated Lightweight Attention ResNet50 (DALAResNet50) model. This model integrates a pre-trained ResNet50 architecture with a lightweight attention mechanism, embedding an attention module in the fourth layer of ResNet50, and incorporates two fully connected layers with LeakyReLU and ReLU activation functions to enhance feature learning capabilities. Extensive experiments conducted on the BreakHis, BACH, and Mini-DDSM datasets demonstrate that DALAResNet50 outperforms state-of-the-art models in accuracy, F1 score, IBA, and GMean, particularly excelling in classification tasks involving imbalanced datasets. Furthermore, the proposed Dynamic Threshold Grad-CAM (DT Grad-CAM) method provides clearer and more focused visualizations, enhancing interpretability and assisting medical experts in identifying key features.

Breast Cancer Classification with Enhanced Interpretability: DALAResNet50 and DT Grad-CAM

TL;DR

This work tackles the challenge of accurate and interpretable breast cancer classification from histopathology images under data imbalance. It introduces DALAResNet50, a ResNet50-based classifier augmented with a lightweight dual-activated attention module, and DT Grad-CAM, a visualization method with adaptive thresholding. Across BreakHis, BACH, and Mini-DDSM, DALAResNet50 achieves superior accuracy, F1, IBA, and GMean compared with eight baselines, while DT Grad-CAM provides clearer, more localized heatmaps. Together, these contributions enhance diagnostic performance and transparency, supporting clinical decision-making and broader medical imaging applications.

Abstract

Automatic classification of breast cancer in histopathology images is crucial for accurate diagnosis and effective treatment planning. Recently, classification methods based on the ResNet architecture have gained prominence due to their ability to improve accuracy significantly. This is achieved by employing skip connections to mitigate vanishing gradient issues, enabling the integration of low-level and high-level feature information. However, the conventional ResNet architecture faces challenges such as data imbalance and limited interpretability, which necessitate cross-domain knowledge and collaboration among medical experts. To address these challenges, this study proposes a novel method for breast cancer classification: the Dual-Activated Lightweight Attention ResNet50 (DALAResNet50) model. This model integrates a pre-trained ResNet50 architecture with a lightweight attention mechanism, embedding an attention module in the fourth layer of ResNet50, and incorporates two fully connected layers with LeakyReLU and ReLU activation functions to enhance feature learning capabilities. Extensive experiments conducted on the BreakHis, BACH, and Mini-DDSM datasets demonstrate that DALAResNet50 outperforms state-of-the-art models in accuracy, F1 score, IBA, and GMean, particularly excelling in classification tasks involving imbalanced datasets. Furthermore, the proposed Dynamic Threshold Grad-CAM (DT Grad-CAM) method provides clearer and more focused visualizations, enhancing interpretability and assisting medical experts in identifying key features.
Paper Structure (17 sections, 9 equations, 7 figures, 14 tables)

This paper contains 17 sections, 9 equations, 7 figures, 14 tables.

Table of Contents

  1. Introduction
  2. Related works
  3. Deep Learning-Based Breast Cancer Image Classification
  4. Grad-CAM-based Visualization Techniques in Histopathology Image Analysis
  5. Methods
  6. ResNet50
  7. Dual-activated lightweight attention ResNet50
  8. Experiment
  9. Implementation Details
  10. Dataset
  11. Data Preprocessing and Splitting
  12. Metrics
  13. Comparison with Other Methods
  14. DT Grad-CAM
  15. DT Grad-CAM Method
  16. ...and 2 more sections

Figures (7)

  • Figure 1: The framework of DALAResNet50 model
  • Figure 2: Visualization of DALAResNet50 model
  • Figure 3: Images of Breast Cancer Histopathology from the BreakHis Dataset at Four Magnifications
  • Figure 4: Images of Breast Cancer Histopathology from the ICIAR2018 Dataset
  • Figure 5: Images of Breast Cancer Histopathology from the Mini-DDSM Dataset
  • ...and 2 more figures