Brain Tumor MRI Classification

This project uses a deep learning model to classify brain tumors from MRI images. The model is a ResNet-18 convolutional neural network that has been pre-trained on the ImageNet dataset and fine-tuned for this specific task. The goal is to accurately classify MRI images into four categories: glioma, meningioma, pituitary, and no tumor.

Kaggle Dataset: https://www.kaggle.com/datasets/masoudnickparvar/brain-tumor-mri-dataset

Key Features

• Transfer Learning: Utilizes a pre-trained ResNet-18 model to leverage learned features from a large dataset, improving performance and reducing training time.
• Data Augmentation: Applies transformations like random resized cropping and horizontal flipping to the training data to increase its diversity and prevent overfitting.
• Early Stopping: Monitors the validation accuracy and stops training if there is no improvement for five consecutive epochs, saving time and computational resources.

Training and Evaluation

The model is trained using the AdamW optimizer and cross-entropy loss with label smoothing. A learning rate scheduler (ReduceLROnPlateau) is used to adjust the learning rate during training. The final evaluation is performed on a separate test set, and the results are visualized with a confusion matrix.

How to Use

1. Set up the environment:

   pip install -r requirements.txt

2. Organize the data: Place the training and testing datasets in the data/ directory, following the structure below:

   data/
   ├── Training/
   │   ├── glioma/
   │   ├── meningioma/
   │   ├── notumor/
   │   └── pituitary/
   └── Testing/
       ├── glioma/
       ├── meningioma/
       ├── notumor/
       └── pituitary/
   

3. Run the notebook: Open and run the notebook.ipynb file in the src/ directory to train the model and see the results.