CNN Visualizations for Medical Dataset

Overview

This repository contains multiple CNN visualization techniques applied to a medical dataset. These visualizations, particularly Grad-CAM and guided backpropagation, are highly effective in interpreting CNNs by highlighting important regions in medical images that influence the network's decisions. They provide valuable insights for medical diagnosis and research.

Contents

• grad-cam-for-kidney-dataset-VGG.ipynb: Grad-CAM visualization using VGG model.
• grad-cam-for-kidney-dataset.ipynb: Grad-CAM visualization.
• guided-backpropogation-alexnet.ipynb: Guided backpropagation using AlexNet.
• simple_CNN_with_Guided_Backpropogation.ipynb: Simple CNN with guided backpropagation.

Requirements

• Python 3.x
• Jupyter Notebook
• Necessary Python libraries (TensorFlow, Keras, etc.)

Usage

1. Clone the repository.
2. Open the Jupyter notebooks.
3. Run the cells to generate visualizations.

Future Work

• Implement other visualization approaches such as saliency maps and occlusion sensitivity.

Authors

• Navaneeth Sivakumar