This website was established as part of an ongoing research effort to provide the medical imaging community with a comprehensive repository of reference standard data sets for objective and rigrorous evaluation of deformable image registration (DIR) spatial accuracy performance. Currently, our data sets consist of:
- Thoracic 4DCT images acquired as part of the standard planning process for the treatment of thoracic malignancies at our institution, The University of Texas M. D. Anderson Cancer Center in Houston, TX, and
- Inspiratory and exipratory breath-hold CT image pairs acquired from the National Heart Lung Blood Institute COPDgene study archive.
Each available image data set has associated with it a coordinate list of anatomical landmarks that have been manually identified and registered by an expert in thoracic imaging, with repeat registration performed by multiple observers to estimate the spatial variance in feature identification. The point sets serve as a reference for evaluating DIR spatial accuracy within the lung for each case. Detailed descriptions of the available images and landmark sets are contained herein.
Investigators are encouraged to register (for free) on this website to access the image volumes for registration with their own DIR algorithms. Optionally, published spatial accuracy results will be posted for comparison across algorithms and implementations. Additionally, we welcome suggested additions to the data set provided here, as well as links to currently available image sets and DIR algorithms online. To register, complete the registration form and email as an attachment to: Inquiries@DIR-lab.com.
Also included for free download is a series of "lite" software packages with application to various aspects of the DIR evaluation process, including manual landmark registration and statistical analysis of measured errors. A series of DIR packages will soon be available, with performance characteristics reported in the Results section of this website.