data

Downloading the data used for the experiments

In this work, we used a large (N=242) dataset for Vestibular Schwannoma segmentation. This dataset is publicly available on TCIA.

This readme explains how to download and pre-process the raw data from TCIA. We also provide an open access to the extreme points and pre-computed geodesics used in this work.

Download the fully annotated TCIA-VS dataset

Option 1 - Downloading the T2 scans only and their segmentation maps (Recommended):

Please follow the following steps:

Step 1: Download the NBIA Data Retriever:

• Please follow the instructions here.

Step 2: Download the T2 scans only:

• Open manifest-T2.tcia with NBIA Data Retriever and download the T2 images (DICOM, 6GB) with the "Descriptive Directory Name" format.

Step 3: DICOM to Nifti conversion:

• Install dependencies: pip install -r preprocess_requirements.txt
• Execute the conversion script: python3 convert.py --input <input_folder> --output <output_folder>
  • <input_folder> is the directory containing the raw T2 images (e.g. /home/admin/manifest-T2/Vestibular-Schwannoma-SEG/).
  • <output_folder> is the directory in which the pre-processed data will be saved.

Step 4: Download the fully annotated segmentation masks here.

Option 2 - Downloading the full dataset and manually convert contours into segmentation masks:

Please follow the instructions from the VS_Seg repository.

Download the extreme points and pre-computed geodesics

The manual and simulated extreme points can be found here. The pre-computed geodesics using the image gradient information (grad folder) and with the additional Euclidean distance (grad_eucl folder) can be found here.

The 6 extreme points are the voxels with the values 1,2,3,4,5,6. Specifically, the pairs of extreme points in the x, y and z axis are respectively {1,2}, {3,4} and {5,6}.

Citations

If you use this VS data, please cite:

@article { ShapeyJNS21,
      author = "Jonathan Shapey and Guotai Wang and Reuben Dorent and Alexis Dimitriadis and Wenqi Li and Ian Paddick and Neil Kitchen and Sotirios Bisdas and Shakeel R. Saeed and Sebastien Ourselin and Robert Bradford and Tom Vercauteren",
      title = "{An artificial intelligence framework for automatic segmentation and volumetry of vestibular schwannomas from contrast-enhanced T1-weighted and high-resolution T2-weighted MRI}",
      journal = "Journal of Neurosurgery JNS",
      year = "2021",
      publisher = "American Association of Neurological Surgeons",
      volume = "134",
      number = "1",
      doi = "10.3171/2019.9.JNS191949",
      pages= "171 - 179",
      url = "https://thejns.org/view/journals/j-neurosurg/134/1/article-p171.xml"
}

If you use the extreme points, please additionally cite:

@article{InExtremIS2021Dorent,
         author={Dorent, Reuben and Joutard, Samuel and Shapey, Jonathan and
         Kujawa, Aaron and Modat, Marc and Ourselin, S\'ebastien and Vercauteren, Tom},
         title={Inter Extreme Points Geodesics for End-to-End Weakly Supervised Image Segmentation},
         journal={MICCAI},
         year={2021},
}

Credits

The conversion script is based on https://github.com/KCL-BMEIS/VS_Seg/tree/master/preprocessing.