Brain Tumor Detection and Segmentation using Machine Learning

🚀 Introduction

Brain tumors are among the most critical health challenges worldwide, with high mortality rates due to late diagnosis and limited access to advanced imaging techniques. Early detection significantly improves survival rates, making AI-driven solutions crucial for faster and more accurate diagnoses.

🧠 What is a Brain Tumor?

A brain tumor is an abnormal growth of cells in the brain. Tumors can be: - Benign (non-cancerous) – Slower growth, less likely to spread. - Malignant (cancerous) – Aggressive, fast-spreading, and life-threatening.

📊 Brain Tumor Statistics

• In Kenya, over 3,000 brain tumor cases are reported annually, with many going undiagnosed due to limited access to specialists.
• Worldwide, brain and nervous system cancers cause nearly 250,000 deaths annually (WHO).
• Early detection can improve survival rates by up to 80% with timely medical intervention.

🌍 Why This Project Matters

This AI-driven project aims to: ✅ Detect brain tumors from MRI scans with high accuracy ✅ Segment tumors for precise medical analysis ✅ Improve healthcare accessibility by automating preliminary diagnoses ✅ Reduce misdiagnoses and assist radiologists in decision-making

🏗️ Project Structure

This project consists of:

1. CNN-Based Model for classification (Tumor vs. No Tumor)
2. Segmentation Model for tumor localization
3. Transfer Learning Implementation to compare performance
4. Flask Web App for user-friendly interaction

🖥️ Technologies Used

• Python, TensorFlow, Keras – Deep Learning Frameworks
• OpenCV, PIL – Image Processing
• Flask – Web Application
• NumPy, Pandas, Matplotlib – Data Handling & Visualization

🔬 Model Details

We use Convolutional Neural Networks (CNNs) to classify brain scans as Tumor or No Tumor. Additionally, U-Net segmentation is applied to detect tumor regions.

Baseline Model (CNN)

• Image Classification Model trained on MRI scan datasets.
• 64x64 input image size, Categorical Cross-Entropy Loss, Adam Optimizer.

Transfer Learning (EfficientNetV2 / ResNet50)

• Pretrained models improve accuracy and reduce training time.

Segmentation Model (U-Net / Mask R-CNN)

• Highlights the tumor region in MRI images for better diagnosis.

🔧 How to Run the Project

1️⃣ Install Dependencies

pip install -r requirements.txt

2️⃣ Train the Model

python train.py

3️⃣ Run Flask App

python app.py

4️⃣ Upload an MRI Scan & Get Prediction!

🚀 Future Improvements

🔹 Enhance segmentation accuracy with advanced architectures
🔹 Deploy as a cloud-based API for hospitals
🔹 Improve dataset size & diversity for generalization
🔹 Integrate explainability (XAI) for medical insights

❤️ Impact in Healthcare

This project demonstrates how AI in radiology can:
✔️ Improve diagnostic speed & accuracy
✔️ Help medical professionals detect tumors early
✔️ Reduce misdiagnoses in low-resource settings
✔️ Save lives through early intervention

---

📌 GitHub Repo: [(https://github.com/LABOSO123)]
✉️ Contact: [labosofaith5@gmail.com]

🚀 Let's revolutionize AI in healthcare!