UW-Madison GI Tract Segmentation Data Tools

Introduction

This repository contains code for loading and visualizing MRI data and annotations used in UW-Madison GI Tract Image Segmentation challenge. See below for a sample visualization.

Changelog

03/21/2023: Initial commit.

Installation

Follow INSTALL.md for installing necessary dependencies

Dataset and Data Preparation

UW-Madison GI Tract Image Segmentation Dataset

The MRI data and their annotations are hosted on Kaggle. A detailed description of the dataset and the downloading link can be found in this webpage.

Note: The training set can be directly downloaded, whereas the test set is only available when the code is submitted to Kaggle. For submission to Kaggle, please refer to instructions here.

Download and Unpack the Dataset

To use the code in this repository, please follow the steps to download and prepare the dataset.

Download uw-madison-gi-tract-image-segmentation.zip from this webpage. A Kaggle account is needed.
Unpack the zip file under ./data (or elsewhere and link to ./data).
The folder structure should look like

This folder
│   README.md
│   ...  
│
└───data/
│    └───train/
│    │	 └───case2
│    │	 └───case6
│    │	 └───...
│    └───train.csv
│   ...
|

Loading MRI Data and Segmentation Masks

To see how to use the code, run

python ./load_data.py --help

To load a single scan (e.g., case136_day25), run

python ./load_data.py ./data/train.csv ./data/train case136_day25

To visualize a single scan (e.g., case136_day25), run

python ./load_data.py ./data/train.csv ./data/train --viz

An animated GIF (e.g., case136_day25.gif) will be saved under ./outputs.

Other Part of the Code

This repository also contains sample code for evaluating 3D Dice score and 3D Hausdorff distance. See common.py.

Contact

Yin Li (yin.li@wisc.edu)

happyharrycn/UW-Madison-GI-Tract-Segmentation-Data-Tools