Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Breast Cancer Diagnosis in Mammography: Evaluation and Integration of Convolutional Neural Networks and Explainable AI

Maryam Ahmed, Tooba Bibi, Rizwan Ahmed Khan, Sidra Nasir

TL;DR

This work addresses the opacity of CNN-based mammography diagnosis by integrating Convolutional Neural Networks with Explainable AI techniques. Using the CBIS-DDSM dataset, it demonstrates that a fine-tuned ResNet50 achieves 0.76 test accuracy after transfer learning and data augmentation, while Grad-CAM, SHAP, and LIME provide complementary explanations evaluated against expert ROIs via the Hausdorff distance. The results indicate that Grad-CAM and SHAP offer more stable and faithful explanations than LIME, thereby enhancing trust and potential clinical adoption. The study also outlines a robust preprocessing and augmentation pipeline and points to future directions in multi-modal data integration and more robust XAI approaches for clinical practice.

Abstract

The Deep learning (DL) models for diagnosing breast cancer from mammographic images often operate as "black boxes", making it difficult for healthcare professionals to trust and understand their decision-making processes. The study presents an integrated framework combining Convolutional Neural Networks (CNNs) and Explainable Artificial Intelligence (XAI) for the enhanced diagnosis of breast cancer using the CBIS-DDSM dataset. The methodology encompasses an elaborate data preprocessing pipeline and advanced data augmentation techniques to counteract dataset limitations and transfer learning using pre-trained networks such as VGG-16, Inception-V3 and ResNet was employed. A focal point of our study is the evaluation of XAI's effectiveness in interpreting model predictions, highlighted by utilizing the Hausdorff measure to assess the alignment between AI-generated explanations and expert annotations quantitatively. This approach is critical for XAI in promoting trustworthiness and ethical fairness in AI-assisted diagnostics. The findings from our research illustrate the effective collaboration between CNNs and XAI in advancing diagnostic methods for breast cancer, thereby facilitating a more seamless integration of advanced AI technologies within clinical settings. By enhancing the interpretability of AI driven decisions, this work lays the groundwork for improved collaboration between AI systems and medical practitioners, ultimately enriching patient care. Furthermore, the implications of our research extended well beyond the current methodologies. It encourages further research into how to combine multimodal data and improve AI explanations to meet the needs of clinical practice.

Enhancing Breast Cancer Diagnosis in Mammography: Evaluation and Integration of Convolutional Neural Networks and Explainable AI

TL;DR

This work addresses the opacity of CNN-based mammography diagnosis by integrating Convolutional Neural Networks with Explainable AI techniques. Using the CBIS-DDSM dataset, it demonstrates that a fine-tuned ResNet50 achieves 0.76 test accuracy after transfer learning and data augmentation, while Grad-CAM, SHAP, and LIME provide complementary explanations evaluated against expert ROIs via the Hausdorff distance. The results indicate that Grad-CAM and SHAP offer more stable and faithful explanations than LIME, thereby enhancing trust and potential clinical adoption. The study also outlines a robust preprocessing and augmentation pipeline and points to future directions in multi-modal data integration and more robust XAI approaches for clinical practice.

Abstract

The Deep learning (DL) models for diagnosing breast cancer from mammographic images often operate as "black boxes", making it difficult for healthcare professionals to trust and understand their decision-making processes. The study presents an integrated framework combining Convolutional Neural Networks (CNNs) and Explainable Artificial Intelligence (XAI) for the enhanced diagnosis of breast cancer using the CBIS-DDSM dataset. The methodology encompasses an elaborate data preprocessing pipeline and advanced data augmentation techniques to counteract dataset limitations and transfer learning using pre-trained networks such as VGG-16, Inception-V3 and ResNet was employed. A focal point of our study is the evaluation of XAI's effectiveness in interpreting model predictions, highlighted by utilizing the Hausdorff measure to assess the alignment between AI-generated explanations and expert annotations quantitatively. This approach is critical for XAI in promoting trustworthiness and ethical fairness in AI-assisted diagnostics. The findings from our research illustrate the effective collaboration between CNNs and XAI in advancing diagnostic methods for breast cancer, thereby facilitating a more seamless integration of advanced AI technologies within clinical settings. By enhancing the interpretability of AI driven decisions, this work lays the groundwork for improved collaboration between AI systems and medical practitioners, ultimately enriching patient care. Furthermore, the implications of our research extended well beyond the current methodologies. It encourages further research into how to combine multimodal data and improve AI explanations to meet the needs of clinical practice.
Paper Structure (21 sections, 2 equations, 14 figures, 2 tables)

This paper contains 21 sections, 2 equations, 14 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Explainability
  3. SHapley Additive exPlanations (SHAP)
  4. Gradient-weighted Class Activation Mapping (Grad-CAM)
  5. Local Interpretable Model-Agnostic Explanations (LIME)
  6. Related Work
  7. Methodology
  8. Dataset
  9. Image Pre-processing
  10. Artifact Reduction
  11. Line Removal
  12. Image Enhancement
  13. Data Augmentation
  14. Transfer Learning
  15. Results
  16. ...and 6 more sections

Figures (14)

  • Figure 1: Breast Cancer Diagnosis system overview using CNNs and XAI.
  • Figure 2: A benign mammogram and its Region of Interest (ROI) on the left are contrasted with a malignant mammogram and its ROI on the right lee2017curated.
  • Figure 3: The five ROI images, proceeding from left to right, are merged into a single comprehensive ROI image on the right.
  • Figure 4: Mammograms displaying unwanted text (on the left) and extraneous objects (on the right).
  • Figure 5: Mammograms featuring distinct lines along their peripheral edges
  • ...and 9 more figures