v2.2.0
Installation
Requirements: Arrayfire, also OpenCL(Windows, on Linux run sudo apt install ocl-icd-opencl-dev) and/or CUDA.
For Python, pip is highly recommended: pip install omegatomo
Note that the examples and MATLAB-files are not included in the pip version.
For MATLAB, the mltbx-file below is recommended. Python files are not included in this version.
pip version and mltbx-file contain prebuilt files for both Windows and Linux. These should generally work, but if not, you can always compile the files yourself.
For Octave, clone or download the source and compile the mex-files. This can be done for Python and/or MATLAB, too. If you want to use both MATLAB and Python, for example, it might be better to download the full source and manually compile the files.
If upgrading, simply overwrite the previous files (if you cloned the files, you'll need recompile the necessary files, too). This should be automatic if using pip. For the mltbx-file, you might end up with two different toolboxes, the old one and the new one. The new one has the name "OMEGA - Open-source multi-dimensional tomographic reconstruction software" (version is 2.2.0), so any other ones should be removed (you should recover any modified example-files you need first though).
What's Changed
For a more complete list of changes see release notes.
Highlights of new features:
- Added temporal regularization and in general improved dynamic reconstruction
- Multi-resolution support for SPECT
- Added Parker weights to FDK, enable with setting
options.useParkerWeightsto true - volume3Dviewer now supports 4D data and works in interactive Python sessions
Merged pull-requests:
Big thanks for these PRs to @saarlemo and @RaziAzizi.
- Dynamic SPECT features by @saarlemo in #74
- Add Barzilai_borwein (BB) method modifications on python by @RaziAzizi in #78
Full Changelog: v2.1.0...v2.2.0
