Brain Tumor Segmentation And Classification using ML

TODO: Better readme

A project to classify and perform segmentation for Brain tumors in Brain MRI images. It is successfully able to classify if a person has a tumor, or not, and locates the tumor if present.

Installation and Usage

This project uses pipenv for dependency management. You need to ensure that you have pipenv installed.

Here are the commands to facilitate using this project.

Clone the repo

git clone https://github.com/Rohith04MVK/Brain-Tumor-Segmentation-And-Classification

Install dependencies and Open the shell

# Install dependencies
pipenv sync -d

# Open the venv shell
pipenv shell

Download data

# Download segmentation data
cd data/segmentation
python download_segmentation_data.py

# Download classification data
cd data/classification
python download_classification_data.py

Run the scripts present

# Run the segmentation script
python src/train_seg.py

# Run the classification script
python src/train_clf.py

Run the main script

python example.py

Note: You can get the pretrained models here.

Project structure

This project has 3 main sections.

• data contains the scripts to download data.
• notebooks contains the well-documented jupyter notebooks.
• src contains the scripts for training and interacting with models.

Data

The dataset being used here is lgg-mri-dataset. This has the brain MRI images, and their respective masks too (as .tif files).

When using, this dataset is split into,

• 3006 Train images
• 590 Testing images
• 333 Validation images

Here is an image of the Mask and the MRI image

This is the mask applied on the MIR

Predictions

The model returns a pandas data frame as its output

These are the model predictions for the images,