svmbir logo

svmbir (super-voxel model-based iterative reconstruction) is an easy-to-use python package for fast iterative reconstruction of tomography data using model-based priors. The svmbir package is a python interface to HPImaging’s optimized C implementation [sv-mbirct] of the super-voxel algorithm [A3] [A4].


  • Supports MBIR reconstruction with Bayesian and Plug-and-Play prior models

  • Supports parallel-beam and fan-beam geometries

System Requirements

  • Python 3.9-3.12

Local builds also require,

  • GCC compiler version 4.8.5 or above

  • OpenMP Libraries

Optional System Requirements

Fastest reconstruction can be obtained with,

  • Intel-based CPU(s) supporting AVX2,AVX512

  • Intel ICC compiler (included with Intel’s free “oneAPI” toolkit)

We also recommend installation to a conda virtual environment.


The project is licensed under the BSD 3-Clause License.

Indices and tables