Multiclass Medical X-ray Image Classification using Deep Learning with Explainable AI

This repository presents a major project focused on the classification of medical X-ray images into multiple categories using deep learning techniques, complemented by Explainable AI (XAI) methods to enhance model interpretability.

Table of Contents

• Project Overview
• Dataset
• Model Architectures
• Explainable AI Techniques
• Results
• Project Structure
• Contributing
• License

Project Overview

The primary objective of this project is to develop a deep learning-based system capable of classifying medical X-ray images into multiple disease categories. Given the critical nature of medical diagnoses, the project also integrates Explainable AI techniques to provide insights into the model's decision-making process, thereby enhancing trust and reliability.

Dataset

The project utilizes a curated dataset of medical X-ray images, encompassing various classes corresponding to different medical conditions. The dataset has been preprocessed and organized for training and evaluation purposes.

Model Architectures

Several deep learning architectures have been explored and implemented in this project:

• ResNet: A deep residual network that addresses the vanishing gradient problem, allowing for the training of very deep networks.
• VGG: A convolutional neural network known for its simplicity and depth, using small convolutional filters.
• Xception: An architecture that relies on depthwise separable convolutions, offering a balance between performance and computational efficiency.

Each model has been trained and evaluated to determine its effectiveness in classifying medical X-ray images.

Explainable AI Techniques

To interpret and visualize the decision-making process of the deep learning models, Explainable AI techniques have been employed. These methods help in highlighting the regions of the X-ray images that the models focus on while making predictions, thereby providing transparency and aiding in clinical validation.

Results

The performance of each model has been evaluated using standard metrics such as accuracy, precision, recall, and F1-score. The comparison.ipynb notebook provides a comprehensive comparison of the models' performances.

Additionally, the integration of Explainable AI techniques offers visual interpretations of the models' predictions, aiding in understanding and validating the results.

Project Structure

The repository is organized as follows:

• Resnet.ipynb: Implementation and training of the ResNet model.
• VGG.ipynb: Implementation and training of the VGG model.
• xception.ipynb: Implementation and training of the Xception model.
• comparison.ipynb: Comparative analysis of all implemented models.
• preprocess.ipynb: Data preprocessing steps.
• Research Paper IJRASET.pdf: Detailed documentation of the project, methodologies, and findings.
• config.yml: Configuration file containing parameters and settings.
• processed/: Directory containing processed data.
• graphs/: Directory containing generated graphs and visualizations.
• ui/: Directory related to the user interface components (if any).

Contributing

Contributions are welcome! If you have suggestions, improvements, or encounter issues, please open an issue or submit a pull request.