README.md

metadata

license: cc-by-nc-sa-4.0
tags:
  - medical-imaging
  - echocardiography
  - 3d-ultrasound
  - cardiac
  - segmentation
task_categories:
  - image-segmentation
pretty_name: CETUS - 3D Echocardiography
size_categories:
  - n<1K

CETUS — Cardiac Echo Training and Ultrasound Segmentation

Description

The CETUS dataset contains 45 patients of real 3D echocardiography volumes acquired at two cardiac phases:

• End-Diastole (ED) — patientXX_ED.nii.gz
• End-Systole (ES) — patientXX_ES.nii.gz

Ground-truth left-ventricular segmentation masks are provided as patientXX_ED_gt.nii.gz and patientXX_ES_gt.nii.gz.

File Structure

dataset/
  patient01/
    patient01_ED.nii.gz
    patient01_ED_gt.nii.gz
    patient01_ES.nii.gz
    patient01_ES_gt.nii.gz
  patient02/
    ...
  ...
  patient45/
    ...
  LICENSE_TERMS.md
  MANDATORY_CITATION.md
jupyter/
  script_cetus.ipynb
  LICENSE_TERMS.md
  MANDATORY_CITATION.md
Resources.zip

Format

• Volume format: NIfTI (.nii.gz)
• Typical shape: varies per patient (see metadata in NIfTI headers)

Usage

import SimpleITK as sitk

img = sitk.ReadImage("dataset/patient01/patient01_ED.nii.gz")
volume = sitk.GetArrayFromImage(img)  # shape: (Z, Y, X)
print(f"Shape: {volume.shape}, dtype: {volume.dtype}")

Source

Original dataset page: https://www.creatis.insa-lyon.fr/Challenge/CETUS/databases.html

License

CC BY-NC-SA 4.0 — free for non-commercial scientific research only.