Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Exploring Explainable AI Techniques for Improved Interpretability in Lung and Colon Cancer Classification

Mukaffi Bin Moin, Fatema Tuj Johora Faria, Swarnajit Saha, Busra Kamal Rafa, Mohammad Shafiul Alam

TL;DR

This work evaluates eight pre-trained CNNs for histopathology-based lung and colon cancer classification using LC25000, incorporating GradCAM, ScoreCAM, LayerCAM, and saliency methods to enhance interpretability. The methodology employs standardized preprocessing and TensorFlow-based training, with hyperparameters tuned per model. Results show exceptionally high accuracy, with Xception achieving 0.9989 and a low 0.0384 log loss, while XAI visualizations provide insight into model decisions. The study highlights the viability of combining state-of-the-art CNNs with explainability tools to support clinical decision-making, while noting limitations in data diversity and the need for multimodal, real-world validation.

Abstract

Lung and colon cancer are serious worldwide health challenges that require early and precise identification to reduce mortality risks. However, diagnosis, which is mostly dependent on histopathologists' competence, presents difficulties and hazards when expertise is insufficient. While diagnostic methods like imaging and blood markers contribute to early detection, histopathology remains the gold standard, although time-consuming and vulnerable to inter-observer mistakes. Limited access to high-end technology further limits patients' ability to receive immediate medical care and diagnosis. Recent advances in deep learning have generated interest in its application to medical imaging analysis, specifically the use of histopathological images to diagnose lung and colon cancer. The goal of this investigation is to use and adapt existing pre-trained CNN-based models, such as Xception, DenseNet201, ResNet101, InceptionV3, DenseNet121, DenseNet169, ResNet152, and InceptionResNetV2, to enhance classification through better augmentation strategies. The results show tremendous progress, with all eight models reaching impressive accuracy ranging from 97% to 99%. Furthermore, attention visualization techniques such as GradCAM, GradCAM++, ScoreCAM, Faster Score-CAM, and LayerCAM, as well as Vanilla Saliency and SmoothGrad, are used to provide insights into the models' classification decisions, thereby improving interpretability and understanding of malignant and benign image classification.

Exploring Explainable AI Techniques for Improved Interpretability in Lung and Colon Cancer Classification

TL;DR

This work evaluates eight pre-trained CNNs for histopathology-based lung and colon cancer classification using LC25000, incorporating GradCAM, ScoreCAM, LayerCAM, and saliency methods to enhance interpretability. The methodology employs standardized preprocessing and TensorFlow-based training, with hyperparameters tuned per model. Results show exceptionally high accuracy, with Xception achieving 0.9989 and a low 0.0384 log loss, while XAI visualizations provide insight into model decisions. The study highlights the viability of combining state-of-the-art CNNs with explainability tools to support clinical decision-making, while noting limitations in data diversity and the need for multimodal, real-world validation.

Abstract

Lung and colon cancer are serious worldwide health challenges that require early and precise identification to reduce mortality risks. However, diagnosis, which is mostly dependent on histopathologists' competence, presents difficulties and hazards when expertise is insufficient. While diagnostic methods like imaging and blood markers contribute to early detection, histopathology remains the gold standard, although time-consuming and vulnerable to inter-observer mistakes. Limited access to high-end technology further limits patients' ability to receive immediate medical care and diagnosis. Recent advances in deep learning have generated interest in its application to medical imaging analysis, specifically the use of histopathological images to diagnose lung and colon cancer. The goal of this investigation is to use and adapt existing pre-trained CNN-based models, such as Xception, DenseNet201, ResNet101, InceptionV3, DenseNet121, DenseNet169, ResNet152, and InceptionResNetV2, to enhance classification through better augmentation strategies. The results show tremendous progress, with all eight models reaching impressive accuracy ranging from 97% to 99%. Furthermore, attention visualization techniques such as GradCAM, GradCAM++, ScoreCAM, Faster Score-CAM, and LayerCAM, as well as Vanilla Saliency and SmoothGrad, are used to provide insights into the models' classification decisions, thereby improving interpretability and understanding of malignant and benign image classification.
Paper Structure (15 sections, 4 figures, 2 tables)

This paper contains 15 sections, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Lung and Colon Cancer Classification without using XAI Techniques
  4. Lung and Colon Cancer Classification using XAI Techniques
  5. Dataset Description
  6. Background Study
  7. Convolutional Neural Networks (CNNs)
  8. Explainable Artificial Intelligence (XAI) techniques
  9. Evaluation Metrics
  10. Proposed Methodology
  11. Results and Discussion
  12. Hyperparameter Settings
  13. Experiments
  14. Limitations and Future Research Directions
  15. Conclusion

Figures (4)

  • Figure 1: Representation of LC25000 dataset
  • Figure 2: Representation of our proposed methodology for lung and colon cancer classification using explainable AI techniques
  • Figure 3: Confusion Matrices of Pretrained CNNs for Lung and Colon Cancer Classification
  • Figure 4: Different versions of AI insights such as GradCAM, GradCAM++, ScoreCAM, Faster ScoreCAM, LayerCAM, Smooth Grad, and Vanilla Saliency provide various explainable views of the original image in the context of lung and colon cancer classification