Stevia-S

Stevia-S / MultiClass-LungDisease-Detection-Using-XAI

Public

Explainable deep learning framework for multi-class lung disease detection from CT scan images using ResNet50, VGG16 feature fusion, and Grad-CAM visualization.

47
1
85% credibility
Found May 17, 2026 at 53 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

This is an educational research project that demonstrates how artificial intelligence can analyze chest CT scans to identify potential lung diseases. The system combines two well-established AI models to study medical images, then uses a technique called Grad-CAM to create colorful heatmaps that show exactly which parts of the scan influenced its decision. Users can upload their own CT images through a web interface and receive both a disease classification (COVID, Normal, or Pneumonia) along with a visual explanation of where the AI focused its attention. The project includes clear disclaimers stating it is for research and educational purposes only, and should not be used for actual medical diagnosis.

How It Works

1
🔬 You discover AI-powered lung analysis

You hear about a tool that can look at chest CT scans and help identify lung diseases like COVID or pneumonia.

2
📋 You learn how it works

The system uses two AI models working together to study your scan, then shows you exactly which parts of the image caught its attention.

3
📤 You upload your CT scan

You drag and drop your chest CT image into the web interface and watch as the system begins its analysis.

4
🧠 The AI studies your image

Behind the scenes, the system examines your scan using deep learning to find patterns associated with lung conditions.

5
You receive your diagnosis
✅
Healthy lungs detected

The AI sees normal patterns and shows mostly blue areas, indicating no concerning regions were found.

⚠️
Potential issue detected

The AI highlights suspicious areas in red and yellow, showing exactly where it spotted concerning patterns.

6
📖 You understand why

The heatmap visualization explains the AI's reasoning, so you can see which parts of your lungs influenced the prediction.

🏥 You have information to share

You now have both a prediction and a visual explanation to discuss with a medical professional if needed.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 53 to 47 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is MultiClass-LungDisease-Detection-Using-XAI?

This is a Python-based medical imaging project that detects lung diseases from CT scans. It combines two pretrained neural networks (ResNet50 and VGG16) into a fusion model that classifies images as COVID, Normal, or Pneumonia. The system generates Grad-CAM heatmaps that highlight which regions of the scan influenced the prediction, making the AI's decision-making transparent. A Streamlit web interface lets users upload CT images and see both the diagnosis and visual explanations side-by-side.

Why is it gaining traction?

The medical AI field needs models that clinicians can actually trust. This project addresses that by combining strong classification performance with visual explanations. The fusion architecture leverages complementary feature extraction from two proven networks, while the web interface makes the technology accessible without coding. The infection severity scoring gives concrete metrics beyond simple classification.

Who should use this?

Researchers exploring explainable AI in medical imaging will find this valuable for understanding how visualization techniques apply to diagnostic models. Developers building proof-of-concept radiology tools can use the Streamlit interface as a starting point. Medical AI students studying transfer learning and CNN interpretability benefit from seeing these concepts applied to real diagnostic tasks.

Verdict

With 47 stars and basic documentation, this is an early-stage educational project rather than production-ready software. The credibility score of 0.85% reflects its nascent community engagement. It is worth exploring for learning purposes, but teams needing robust medical imaging solutions should look elsewhere.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.