TumorVision

Computer Vision Project - Patch to Graph Embedding Framework for Brain Tumor Classification

---

Problem Statement

To develop a hybrid deep learning framework that leverages Graph Neural Networks (GNNs) for effective brain tumor detection and classification from MRI images by capturing both spatial and structural relationships among brain regions.

---

Objectives

• Integrate CNN-based feature extraction with Graph Neural Network (GNN) modeling.
• Convert MRI image patches into graph representations to capture spatial and structural dependencies among brain regions.
• Develop a robust, interpretable, and high-accuracy brain tumor detection and classification model across diverse MRI scans.

---

Tools and Technologies

• Python
• Google Colab
• PyTorch
• OpenCV
• Matplotlib
• Streamlit

---

Dataset

Dataset Name: Brain MRI Images for Brain Tumor Detection Source: Kaggle Dataset Link

• 98 (82 unique) images without tumor
• 155 (140 unique) images with tumor

Note: Download the dataset folders named yes and no from the above Kaggle link. Place them inside the folder brain_tumor_dataset/ in your project directory as shown below:

TUMORVISION/
│
├── brain_tumor_dataset/
│   ├── yes/
│   └── no/

---

Pipeline Overview

1. Data Acquisition and Preprocessing

   • Load and resize MRI images
   • Normalize pixel values

2. Data Augmentation

   • Apply rotation, flipping, and scaling to increase dataset diversity

3. Patch Generation Module

   • Divide MRI images into smaller patches to capture local information

4. CNN-based Feature Extraction Module

   • Extract spatial features from patches using a CNN backbone

5. Graph Construction and Embedding Module

   • Represent each patch as a node
   • Build adjacency matrix based on spatial proximity
   • Generate graph embeddings

6. Graph Neural Network (GNN) Classification Module

   • Classify the graph representations into Tumor / Non-Tumor categories

7. Evaluation and Visualization Module

   • Evaluate model performance using accuracy, precision, recall, F1-score
   • Visualize results using Matplotlib and Streamlit

---

Contributors

Team Members: Jenisa Merlin D, Harithra R