← Homepage

Download

You can download the scribble annotations at this link. File names are the same as in the ACDC dataset. For example, for patient 12 we have the files patient012_frame_ED_scribble.nii.gz and patient012_frame_ES_scribble.nii.gz for the end-diastolic and end-systolic cardiac phases.

If you use this data, you should cite our paper as:

@ARTICLE{9389796,
  author={Valvano, Gabriele and Leo, Andrea and Tsaftaris, Sotirios A.},
  journal={IEEE Transactions on Medical Imaging}, 
  title={Learning to Segment from Scribbles using Multi-scale Adversarial Attention Gates}, 
  year={2021},
  volume={},
  number={},
  pages={1-1},
  doi={10.1109/TMI.2021.3069634}
}

Data Description

ACDC

The 2017 Automatic Cardiac Diagnosis Challenge dataset [1] contains cine-MR images obtained by 100 patients with different MR scanners and acquisition protocols. Manual segmentations are provided along with the images, containing pixel-wise annotations for the end-diastolic (ED) and end-systolic (ES) cardiac phases. The annotated structures are left ventricle (LV), right ventricle (RV) and myocardium (MYO). You can download the data here.

Scribble Generation

To annotate the images with scribbles we used ITK-SNAP [2]. We manually drew scribbles for RV, MYO, LV on top of the available segmentation masks provided in ACDC, for ES and ED phases. We additionally drew a scribble approximately around the heart to identify pixels belonging to the background class (BGD), while leaving the rest of the pixels unlabeled. The average (standard deviation) image coverage of scribbles is: 0.1 (0.1)%, 0.2 (0.1)%, 0.1 (0.1)% and 10.4 (8.4)%, for RV, MYO, LV and BGD, respectively.

Reference

[1] O. Bernard, A. Lalande, C. Zotti, F. Cervenansky, X. Yang, P.-A. Heng, I. Cetin, K. Lekadir, O. Camara, M. A. G. Ballester et al., “Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: Is the problem solved?” IEEE TMI, vol. 37, no. 11, pp. 2514– 2525, 2018.

[2] P. A. Yushkevich, J. Piven, H. Cody Hazlett, R. Gimpel Smith, S. Ho, J. C. Gee, and G. Gerig, “User-Guided 3D Active Contour Segmentation of Anatomical Structures: Significantly Improved Efficiency and Reliability” Neuroimage, vol. 31, no. 3, pp. 1116–1128, 2006.

Don’t miss any update!

You can either: