🧠 Brain Tumor Detection using CNN

🔗 Main Libraries & Tools

• – Deep learning framework for model building
• – High-level API for TensorFlow
• – Image processing and computer vision
• – Visualization library for plots & images
• – Data splitting and evaluation metrics
• – Helper functions for image processing

---

📂 Project Structure

BRAIN_TUMOR_DETECTION/
│── brain_tumor_dataset/
│   ├── no/                # MRI images without tumor
│   └── yes/               # MRI images with tumor
│── cnn-parameters-improvement-24-0.86.model   # Trained CNN weights
│── data_aug.py            # Script for data augmentation
│── final_rslt.py          # Run inference on a single MRI image
│── ver1_train.py          # Train & evaluate CNN model

---

🛠️ Requirements

Install dependencies with:

pip install -r requirements.txt

requirements.txt

tensorflow
numpy
matplotlib
opencv-python
imutils
scikit-learn

---

🚀 How to Use

1. Prepare Dataset
   Organize your data as:

   brain_tumor_dataset/
   ├── no/
   └── yes/
   

2. Augment Data (optional)

   python data_aug.py

3. Train Model

   python ver1_train.py

   This will generate cnn-parameters-improvement-24-0.86.model.
4. Run Inference on a New Image
   Edit final_rslt.py to point to your test image, then run:

   python final_rslt.py

---

📊 Model Performance (from paper)

Accuracy (Test): 0.95  
F1 Score (Test): 0.93

---

🧮 How It Works

• Preprocessing: grayscale conversion → Gaussian blur → thresholding → morphological cleanup → contour cropping.
• CNN Model: convolutional layers → batch norm & pooling → dense classification head.
• Prediction Logic: outputs probability; if > 0.6 → “Brain Tumor Detected,” else “Normal.”

---

📜 Reference & Citation

Please cite our peer-reviewed work if you use this repository:

Brain Tumour Detection Using Deep Learning
ResearchGate Publication

---