README.md

metadata

pretty_name: MSD Cardiac (Task02_Heart)
language:
  - en
license: cc-by-sa-4.0
tags:
  - medical
  - mri
  - cardiac
  - segmentation
  - left-atrium
  - nifti
task_categories:
  - image-segmentation
size_categories:
  - 10-100
configs:
  - config_name: default
    data_files:
      - split: train
        path: train.csv
      - split: test
        path: test.csv

MSD Cardiac — Task02_Heart (Left Atrium Segmentation)

Processed NIfTI data from the Medical Segmentation Decathlon Task02 (Heart). The goal is to segment the left atrium from mono-modal MR images.

Dataset Summary

• Modality: MRI
• Task: Left atrium segmentation
• Patients: 30 total (20 train, 10 test)
• Labels: 0 = background, 1 = left atrium
• Splits: train (with labels), test (images only, no public labels)

Data Structure (per patient)

Each patient directory contains:

• <pid>.nii.gz — MR image volume
• <pid>_gt.nii.gz — segmentation mask (train only)

Columns

Column	Type	Description
pid	string	Patient ID (e.g., la_003)
image	string	Relative path to MR image
label	string	Relative path to segmentation mask (None for test)
orig_spacing_x	float	Original X spacing (mm)
orig_spacing_y	float	Original Y spacing (mm)
orig_spacing_z	float	Original Z spacing (mm)
n_slices	int	Number of slices after resampling
la_volume_cm3	float	Left atrium volume (cm³, train only)
la_proportion	float	Left atrium voxel proportion (train only)

Resolution Details

Statistic	Spacing (mm)	Size
min	(1.25, 1.25, 1.37)	(320, 320, 90)
median	(1.25, 1.25, 1.37)	(320, 320, 115)
max	(1.25, 1.25, 1.37)	(320, 320, 130)

Usage

import pandas as pd
import nibabel as nib

df = pd.read_csv("train.csv")
row = df.iloc[0]
img = nib.load(row["image"])
arr = img.get_fdata()

Source

Official MSD website: http://medicaldecathlon.com/

License

CC-BY-SA 4.0

Citation

@article{antonelli2022medical,
  title={The Medical Segmentation Decathlon},
  author={Antonelli, Michela and Reinke, Annika and Bakas, Spyridon and others},
  journal={Nature Communications},
  year={2022},
  doi={10.1038/s41467-022-30695-9}
}