SIIM-ACR Pneumothorax segmentation

Purpose

This repository attempts to perform segmentation task on Pneumothorax disease from chest x-ray image

Dataset

Pneumothorax can be caused by a blunt chest injury, damage from underlying lung disease, or most horrifying—it may occur for no obvious reason at all. On some occasions, a collapsed lung can be a life-threatening event. Pneumothorax is usually diagnosed by a radiologist on a chest x-ray, and can sometimes be very difficult to confirm. To learn more about dataset visit kaggle website https://www.kaggle.com/c/siim-acr-pneumothorax-segmentation. Since this dataset is no longer on the official competition, you can download the dataset at this site: https://www.kaggle.com/datasets/jesperdramsch/siim-acr-pneumothorax-segmentation-data

Installation

Clone the repository

git clone https://github.com/nguyenbanhducA1K51/Siim-segmentation.git

cd to the repo and use the package manager pip to install libary and package in file requirements.txt (recommend install in conda environment).

pip install -r requirements.txt

Usage

• Open the ../config/config.json file and change the value of following variable: "repo_path": path to the cloned repo

"train_dicom_path": path to the train dicome file, for example "/root/data/siim-other/pneumothorax/dicom-images-train"

"infer_dicom_path": path to the infer dicome file,

"train_rle_csv": path to the train-rle csv file, for exmample "/root/data/siim-other/pneumothorax/train-rle.csv"

"png_save_path": path that will convert dicom files to png files

• First , run the preprocess file at "..src/preprocess.py"

python3 ./src/preprocess.py

• Run the training pipeline by

bash /scripts/train.sh