Techn. Fakultät Website deprecated and outdated. Click here for the new site. FAU-Logo


What is RabbitCT about?
Fast 3-D cone beam reconstruction is mandatory for many clinical workflows. Therefore researchers and industry work hard on hardware-optimized 3-D reconstruction.
Backprojection is a major component of many reconstruction algorithms. It requires a projection of each voxel onto the projection data, and then data interpolation, before updating the voxel value. This step is the bottleneck of most reconstruction algorithms and has been the focus of optimization in recent publications.
A crucial limitation, however, of those publications is that the presented results are not comparable, mainly due to variations in data acquisition, preprocessing, chosen geometries, and the lack of a common publicly available test dataset. We provide an open platform for worldwide comparison in backprojection performance and ranking on different architectures using one specific, clinical, high resolution C-arm CT dataset of a rabbit. This includes a sophisticated benchmark interface, a prototype implementation in C++, and image quality measures.

RabbitCT is a collaboration of the Department of Neuroradiology and the Pattern Recognition Lab at the Friedrich-Alexander-University Erlangen-Nuremberg.

RabbitCT News
You will find up-to-date news on RabbitCT's project page.

Ranking and Submission
Different algorithms and algorithm results can be submitted to our website. The results are clustered by the problem size (256, 512 or 1024; 128 is not considered in the ranking) denoting the corner length of the reconstructed volume in voxels. The results for each problem size are each shown in separate rankings:
  • Ranking for problem size 256 (easy)
  • Ranking for problem size 512 (hard)
  • Ranking for problem size 1024 (only for real rabbits!)
  • Submit your algorithm here.

Where is the Rabbit?
The dataset we provide for reconstruction was acquired from a real rabbit with a real C-arm system. Therefore it can be regarded as clinically relevant. This video shows a volume rendering of our reference reconstruction.

You can view the reconstructed result of the reference implementation in 256x256x256 here. (Java required, loooong loading time (ca. 64 MB).)