Skip to content


Repository files navigation

Validation Data for Deformable Image Registration of the Lungs


We believe research on deformable image registration (DIR) of the lungs will greatly benefit from making image and validation data publicly available. Several 4D CT data sets, acquired at the Léon Bérard Cancer Center, can be downloaded here for use in your research. Along with the images, we provide large sets of manually identified landmarks, allowing you to validate your motion estimation.

We welcome all suggestions that can make this data more useful for you. David Sarrut, david.sarrut[at]

Data Sets

DIR Validation Data

This data set includes:

  • Six 4D CT data sets
  • 100 landmarks in the inhale and exhale frame.
  • For 3 patients, landmarks are also available in all intermediate frames.

This data is fully described in

  • J. Vandemeulebroucke, S. Rit, J. Kybic, P. Clarysse, and D. Sarrut.

    Spatiotemporal motion estimation for respiratory-correlated imaging of the lungs.

    In Med Phys, 2011, 38(1), 166-178. pdf

The inhale-exhale image pairs were also used in

  • J. Vandemeulebroucke, O. Bernard, S. Rit, J. Kybic, P. Clarysse, and D. Sarrut.

    Automated segmentation of a motion mask to preserve sliding motion in deformable registration of thoracic CT.

    In Med Phys, 2012, 39(2), 1006-1015. pdf

Proceed to Download Page

The POPI-model, a Point-validated Pixel-based Breathing Thorax Model

This data set includes:

  • One 4D CT data set
  • 40 landmarks in all frames.
  • The deformation vector fields (DVF) were obtained using two registration methods (expressed in voxel units, so you need to x2 to get in mm).

This data set was originally described in

  • J. Vandemeulebroucke, D. Sarrut, and P. Clarysse

    The POPI-model, a point-validated pixel-based breathing thorax model.

    In Proceeding of the XVth ICCR Conference, Toronto, Canada. 2007. pdf

Proceed to Download Page

Other Useful Links