Datasets:

autoencodix
/

OrganMNIST3D-subsample

license: cc-by-nc-4.0
pretty_name: OrganMNIST3D Subset (Bladder and Femur-Right)
task_categories:
  - image-classification
tags:
  - biomedical-imaging
  - 3d
  - NIfTI file format (.nii.gz)
  - medmnist+
  - autoencodix

Info

Subset of the MedMNIST+ OrganMNIST3D dataset containing only the organ classes bladder and femur-right, prepared for use with the AUTOENCODIX Imagix3D tutorial. The complete MedMNIST dataset collections can be found here.

OrganMNIST3D is a three-dimensional biomedical image classification dataset derived from abdominal CT scans. The MedMNIST+ version provides the volumes at a spatial resolution of 64 × 64 × 64 voxels.

For this tutorial dataset, only samples belonging to the bladder and right-femur classes were retained. The volumes were converted from its original .npz file format to compressed NIfTI files (.nii.gz).

345 images (174 bladder, 171 femur-right) with a spatial size of 64 × 64 × 64 voxels.
organmnist3d_images.zip: compressed archive containing the individual volumes in NIfTI format (.nii.gz).
organmnist3d_mappings.csv: comma-separated annotation file containing the sample identifiers, paths to the corresponding NIfTI files, and organ-class labels.

Loading

The files can be downloaded from Hugging Face using huggingface_hub:

from huggingface_hub import hf_hub_download
import zipfile

mapping_path = hf_hub_download(
repo_id="autoencodix/OrganMNIST3D-subsample",
repo_type="dataset",
filename="organmnist3d_mappings.csv",
)

zip_path = hf_hub_download(
repo_id="autoencodix/OrganMNIST3D-subsample",
repo_type="dataset",
filename="organmnist3d_images.zip",
)

with zipfile.ZipFile(zip_path) as zf:
zf.extractall("data/images/")

License

Original dataset licensed under Creative Commons Attribution 4.0 International (CC BY 4.0), see: https://medmnist.com/. The Apache-2.0 license associated with the MedMNIST project applies to the MedMNIST software code, see: https://github.com/MedMNIST/MedMNIST/.

Info

Contents

Loading

License