Brain Cancer Classifier and Segmentation

Datasets

The datasets used in this project are available on Kaggle:

• Brain Tumor Image Dataset: Semantic Segmentation
• Brain Tumor MRI Dataset
• Brain MRI Segmentation Dataset

Setup Instructions

Step 1: Install Dependencies

Install the necessary Python packages using the requirements.txt file:

pip install -r requirements.txt

Alternatively, you can create a conda environment using the following commands:

conda env create -f environment.yaml
conda activate ai_project_env

Step 2: Setting Up Pre-commit Hooks

Our project uses pre-commit hooks to ensure the cleanliness and consistency of Jupyter notebooks by automatically stripping outputs before they are committed. This step helps maintain a clean git history and minimizes "diff noise."

After installing the project dependencies, activate the pre-commit hooks by running the following command:

pre-commit install

This command sets up the hooks based on our project's .pre-commit-config.yaml configuration and needs to be run only once.

This current hook cleans the Jupyter notebooks before they are committed.

Step 3: Setup Environment Variables

To create a base configuration for the project, run the following command:

cp config/env_local.env .env

This will create a .env file in the root dir of the project. However, to actually run training and testing scripts, you will need to fill in the values in the .env file.

Step 4: Kaggle API Authentication

Follow the instructions to set up your Kaggle API credentials. You can find the Kaggle API authentication instructions in the Kaggle API Documentation.

Step 5: Download Datasets

Refer to the notebooks/downloading_datasets.ipynb notebook for step-by-step instructions on using the Kaggle API to download the datasets required for this project. The datasets will be downloaded to the ./datasets folder, which is configured to be ignored by git.

Loading Classification Dataset

For an example of how to load the classification dataset, see the notebooks/dataloader_example.ipynb notebook. This notebook demonstrates how to use the TumorClassificationDataset class to load either the Training or Testing split from the Tumor Classification Dataset.

Run Experiments:

Classification

• Logistic Regression - see notebooks/logistic_regression.ipynb
• CNN - see notebooks/cnn_classification.ipynb
• ResNet (binary) - see notebooks/resnet.ipynb
• ResNet (multi) - see notebooks/resnet_multi.ipynb

Semantic Segmentation

See src/scripts/train_segmentation.py for logic related to running segmentation experiments. For more info run the following from the root directory to see available training configurations:

python -m src.scripts.train_segmentation --help

Object Detection

Navigate to the main function in src.scripts.train_object_detection.py and edit the training_configs. To run the specified experiment you can use...

python -m src.scripts.train_object_detection