Thickness calculation on binary 3D images
Project description
Compute the thickness of a solid using Yezzi and Prince method described in the article “An Eulerian PDE Approach for Computing Tissue Thickness”, IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 22, NO. 10, OCTOBER 2003. [1]
A C implementation by Rubén Cárdenes [2] helped me a lot writing this, especially the anisotropic part.
Requirements
numpy, cython, scikit-image. Tested with Debian Jessie and Fedora 27, miniconda-python 3.6, cython 0.24, numpy 1.11.2, scikit image 0.12.3
Installation instruction
Available on pypi. [3] Use pip: pip install pyezzi
Alternatively, clone the repository and build cython modules with python setup.py build_ext --inplace.
Usage
from pyezzi import compute_thickness
thickness = compute_thickness(labeled_image, debug=True)labeled_image is a 3 dimensional numpy array where the wall is labeled 2 and the interior is labeled 1.
A spacing parameter specifying the spacing between voxels along the axes can optionnaly be specified.
Check out the included jupyter notebooks in the example folder for more details.
Note on thickness solver implementation
The ordered traversal method mentioned in the original publication can be used using the yezzi_solver='ordered' keyword argument. However, we found that it introduces artifacts to the result. Also the implementation is in pure python so it is slower to solve than the iterative algorithm.
Contributions
Feel free to submit pull requests. I know the code is nowhere near optimal as it is.
License
This work is licensed under the french CeCILL license. [4] You’re free to use and modify the code, but please cite the original paper and me.
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
