COVID Detection from CXR Using Explainable CNN

[inkerton]

Pull Request for PyVerse 💡

Requesting to submit a pull request to the PyVerse repository.

---

Issue Title

Please enter the title of the issue related to your pull request.
COVID Detection from CXR Using Explainable CNN

• I have provided the issue title.

---

Info about the Related Issue

What's the goal of the project?
Project Description
This project aims to develop a robust and explainable Convolutional Neural Network (CNN) model to accurately detect COVID-19 infections from Chest X-ray (CXR) images. By leveraging the power of deep learning and explainable AI techniques, this model will not only provide accurate predictions but also offer insights into the decision-making process, enhancing trust and transparency in medical diagnosis.

Key Objectives:

Accurate COVID-19 Detection: Develop a highly accurate CNN model capable of differentiating between COVID-19 positive and negative CXR images.
Explainable AI: Implement techniques to visualize and interpret the model's decision-making process, providing insights into the features that contribute to the classification.
Robustness and Generalizability: Ensure the model's robustness by training it on a diverse dataset and evaluating its performance on unseen data.
User-Friendly Interface: Create a user-friendly interface for medical professionals to easily input CXR images and receive accurate predictions with explanations.
Methodology:

Data Acquisition and Preprocessing:

Collect a large and diverse dataset of CXR images, including both COVID-19 positive and negative cases.
Preprocess the images to ensure consistency in size, format, and intensity levels.
Augment the dataset using techniques like rotation, flipping, and noise addition to improve the model's generalization ability.
Model Architecture:

Design a deep CNN architecture, such as VGG16 or ResNet, to extract relevant features from the CXR images.
Incorporate attention mechanisms or other explainable AI techniques to highlight the regions of interest in the images that influence the model's predictions.
Training and Optimization:

Train the model using an appropriate loss function (e.g., categorical cross-entropy) and optimizer (e.g., Adam).
Implement techniques like early stopping and learning rate reduction to prevent overfitting and improve convergence.
Evaluation and Validation:

Evaluate the model's performance using metrics like accuracy, precision, recall, F1-score, and AUC-ROC curve.
Conduct cross-validation to assess the model's generalization ability on different data splits.
Explainability Techniques:

Employ techniques like Grad-CAM, SHAP, or LIME to visualize the model's decision-making process and identify the most influential features.
Generate heatmaps to highlight the regions of the CXR image that contribute most to the classification.

• I have described the aim of the project.

---

Name

Please mention your name.
inkerton

• I have provided my name.

---

GitHub ID

Please mention your GitHub ID.
inkerton

• I have provided my GitHub ID.

---

Email ID

Please mention your email ID for further communication.
[email protected]

• I have provided my email ID.

---

Identify Yourself

Mention in which program you are contributing (e.g., WoB, GSSOC, SSOC, SWOC).
GSSOC

• I have mentioned my participant role.

---

Closes

Enter the issue number that will be closed through this PR.
Closes: #issue-number #1084

• I have provided the issue number.

---

Type of Change

Select the type of change:

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Code style update (formatting, local variables)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

---

Checklist

Please confirm the following:

• My code follows the guidelines of this project.
• I have performed a self-review of my own code.
• I have commented my code, particularly wherever it was hard to understand.
• I have made corresponding changes to the documentation.
• My changes generate no new warnings.
• I have added things that prove my fix is effective or that my feature works.
• Any dependent changes have been merged and published in downstream modules.

[github-actions]

👋 Thank you for opening this pull request! We're excited to review your contribution. Please give us a moment, and we'll get back to you shortly!

Feel free to join our community on Discord to discuss more!

[ruhi47]

Looks good!

[TheChaoticor]

Approved