This project uses VIT and VGG16 for Liver CT Scan image classification.
🩺 Liver Disease Detection using VGG16 and Vision Transformer (ViT) 🎯 A Deep Learning-powered system for automatic liver disease detection from CT scan images, using state-of-the-art VGG16 and ViT architectures.
📌 Project Overview This project implements a web-based application that accepts CT-scan images and predicts the type of liver disease present using two powerful deep learning models: ✅ VGG16 – for feature-rich CNN-based detection ✅ Vision Transformer (ViT) – for advanced attention-based analysis
🧪 Detected Disease Categories The system can classify images into the following categories:
🧬 Carcinoma
🧬 Squamous Cell Carcinoma
🧬 Adenocarcinoma
✅ Normal
🚀 Features 📤 Upload CT Scan image (PNG/JPG)
🧠 Model prediction using pretrained VGG16 / ViT
📊 Disease confidence score
📄 Auto-generated diagnosis report with treatment suggestions
🔥 Heatmap visualization (CAM)
🔐 Secure login system for access control
🛠️ Technologies Used Tech Stack Description 🧠 Deep Learning VGG16, ViT (Vision Transformer) 🐍 Python Core programming language ⚙️ Flask Lightweight web framework 🎨 HTML/CSS For web interface (Bootstrap 5 styled) 🖼️ OpenCV / PIL Image pre-processing 🧪 TensorFlow/Keras For building and loading DL models
📂 Folder Structure csharp Copy Edit . ├── Models/ # Saved VGG16 and ViT models ├── uploads/ # Uploaded CT scan images ├── templates/ │ ├── index.html # Upload & results dashboard │ ├── login.html # Secure login page │ ├── report.html # Diagnosis report page │ └── heat.html # Heatmap page ├── static/ # CSS / Images (if needed) ├── main.py # Flask backend ├── README.md # You're reading it! 💻 How to Run Locally Clone the repository
bash Copy Edit git clone https://github.com/your-username/liver-detection-vgg16-vit.git cd liver-detection-vgg16-vit Install dependencies
bash Copy Edit pip install -r requirements.txt Place your trained model in the Models/ folder as model.h5
Run the server
bash Copy Edit python main.py Visit the app Navigate to http://127.0.0.1:5000 in your browser.
📊 Example Output ✅ Predicted Disease: Adenocarcinoma
📈 Confidence Score: 91.24%
🩺 Recommendations:
Immunotherapy in conjunction with chemotherapy
Monitor tumor markers like AFP regularly
Liver transplant evaluation if condition worsens
🧠 About the Models 🔹 VGG16 16-layer CNN pre-trained on ImageNet
Fine-tuned on liver CT scan dataset
Excellent at extracting visual features like edges, texture, and shape
🔹 Vision Transformer (ViT) Breaks image into patches and uses Transformer encoders
Learns spatial and semantic features via attention
Offers higher performance in some medical tasks due to its long-range attention
🌍 Real-World Applications 🏥 Assist doctors in early liver disease diagnosis
🔍 Reduce human error in medical image analysis
📄 Automatic report generation for patient use
📊 Medical AI research and education
🔐 Login Credentials (for demo) Username: admin
Password: rajdeep
🧾 License This project is licensed under the MIT License. Feel free to fork and enhance it for academic or non-commercial purposes.
🤝 Acknowledgements Kaggle – For dataset resources
Keras Applications – For pretrained models