Brain Tumor MRI Segmentation Project

This project aims to develop a system for detecting and segmenting brain tumors from MRI images. The system uses a ResNet model to classify MRI images as having a tumor or not, and a U-Net model to segment the tumor if it is detected. The frontend is implemented using React.

Download Data

1. Download the brain tumor MRI dataset from Kaggle 'https://www.kaggle.com/datasets/masoudnickparvar/brain-tumor-mri-dataset/data'
2. Create a new folder named data in the root directory of the project.
3. Extract the downloaded dataset into the data folder.

Data Preprocessing

Run the data_preprocessing.py script to preprocess the data.

Generate Masks

Run the unet_model.py script to build the U-Net model, then generate masks using the masks.py script.

Train ResNet Model

Run the train_resnet.py script to train the ResNet model. This script trains the ResNet model on the preprocessed data and saves the trained model in the models/resnet folder.

Train U-Net Model

Run the train_unet.py script to train the U-Net model. This script trains the U-Net model on the preprocessed data and saves the trained model in the models/unet folder.

Inference

Run the inference.py script to start the inference server. This script starts a Flask server that uses the trained ResNet and U-Net models to classify and segment brain tumors in MRI images.

Frontend Implementation

To start the React frontend:

1. Open a terminal and navigate to the react_app directory.
2. Start the React app using npm start.

Ensure that all necessary dependencies are installed, and follow the steps in each section to complete the project setup.