[data request] SIIM-ACR Pneumothorax Segmentation Dataset

• Name of dataset: SIIM-ACR Pneumothorax Segmentation Dataset

• URL of dataset:

  • SIIM-ACR Pneumothorax Segmentation

  • SIIM_TRAIN_TEST

• License of dataset: Open source, any purpose, including commercial use

• Short description of dataset and use case(s):

  • This dataset comes from the SIIM-ACR Pneumothorax Segmentation, held by SIIM in 2019. It contains over 15,000 chest X-ray scans(~12,100 for training, ~3,100 for testing) and corresponding masks for pneumothorax diagnosis segmentation. All scans are stored in DICOM format, which consists of patient demographic information(de-identified) and scan metadata.
    Here's the link to the competition: SIIM-ACR Pneumothorax Segmentation

• This request is part of the first assignment about open source contribution from BME590 Deep Learning in Healthcare at Duke University. Matthew Lin will be working on coding scripts for this dataset along with TFDS guide. For those who are interested in adding this dataset to TFDS, please reply to this issue and I will be happy to collaborate!

Folks who would also like to see this dataset in tensorflow/datasets, please thumbs-up so the developers can know which requests to prioritize.

And if you'd like to contribute the dataset (thank you!), see guide to adding a dataset.

Start with the linked guide to adding a dataset and inspect the SIIM-ACR Pneumothorax Segmentation data on Kaggle, including its DICOM scans and masks. Done means the dataset is added to TensorFlow Datasets with the requested metadata and usable training and testing data; the issue notes that Matthew Lin may already be working on scripts.