This repository provides necessary functions and examples on how to do multi-grid reconstruction on projection image collected in open CT format. For numerical shepp-logan examples, see odl-multigrid.
The script rely on several python modules, paritculatrly
odl
numpy
matplotlib
pickle
os
glob
dicom
The dicom
module can be most easily installed by easy_install pydicom
. Note that the conversion here intends to generate DICOM-CT-PD-format, instead of standard DICOM. For this, the default _dicom_dict.py
dictionary in pydicom
has to be replaced by the _dicom_dict.py
file given in this repository. Generally, this file (to be replaced) can be found in /path/to/lib/python2.7/site-packages/pydicom-0.9.8-py2.7.egg/dicom/
.
For the reconstruction, the python library odl
is used.
Function to save reconstructed data file as pickled .txt-files
Function to display multigrid reconstructions, overlaying fine and coarse grid reconstructions. The function is created for one ROI, and several ROI display remains to be implemented. Note that for the dual-display, the coarse grid is expanded (using ExpandMatrix
) to have the same artificial discretization as the fine grid.
display_function
also has a keword argument crossline
that enables plotting over a defined line in the middle of the reconstruction ROI.
Generate sinogram from input openCT dicom-files. Option input in the form of light field and potential min/max definitions for truncation.
Modified odl-scripts, basically created to enable output of time/iteration and the L2-norm between Radon transform forward operator and projection data. For more information see functions/odl_modification/README.md
FBP example.
CG example.
TV example.
CG example with 2 fine-discretization ROIs. Note that the display function cannot handle this case for now.