Segmentator Versions Save

Minor maintenance release fixing nibabel, numpy, and matplotlib deprecations errors and warnings.

Major changes:

• GUI elements now render properly also in Matplotlib version 3.x.
• License is changed to BSD-3-Clause.
• Normalized graph cut feature is deprecated because of dependency problems. Older versions of Segmentator can still be used in Python 2 and following the instructions on the related wiki page. This feature might be re-implemented from scratch in the future if I have the resources.

Minor changes:

• Fixed nifti header being not copied properly from the source image when using segmentator_filters.
• Tested and updated other dependencies (numpy, scipy) to more recent versions in setup.py and requirements.txt.

New features in version 1.5:

• Lasso drawing mode erase function is added to GUI.
• Deriche filter for determining the smoothness of the gradients. Useful for very high resolution data.
• Non-linear anisotropic diffusion based smoothing filters (Weickert 1998) are added as a new utility accessible through command line:
  • segmentator_filters -h

Fixes in the current version:

• [1.5.2] Minor bugs related to gradient magnitude image exports.
• [1.5.2] Updated links after publication of the paper and some readme cleaning.
• [1.5.3] Fixed module not found error in segmentator_filters

New features:

• Lasso drawing mode erase function is added to GUI.
• Deriche filter for determining the smoothness of the gradients. Useful for very high resolution data.
• Non-linear anisotropic diffusion based smoothing filters (Weickert 1998) are added as a new utility accessible through command line:
  • segmentator_filters -h

New features

• Export gradient magnitude flag added: --export_gramag
• Excluding zero voxels by default from histogram and labeling process. Can be turned off by using: --include_zeros

For users:

• New gradient magnitude operators added (Scharr, Sobel, Prewitt).
• The default gradient magnitude calculation is switched to 3D Scharr.
• Switched to Matplotlib 2.0 (Matplotlib 1.5.3 is not compatible anymore).
• Rotation button added.
• New options in the command line interface.
• Small improvements in N-cut pipeline.

For developers:

• Travis and Appveyor continuous integration added.
• Started writing tests :).

Major reorganization of the project structure.

• New terminal interface, see the latest installation instructions in our wiki.
• Default parameters are tweaked for the ncut feature.
• Most functions are now script-able, new wiki pages with examples will follow soon (2017 Summer).

There are some big changes in this version:

1. Segmentator can be called from the command line.
2. Many GUI improvements & additional controls.
3. Highlighter for the bin of the selected voxel in the image browser (data probe).
4. Experimental features (Deriche gradient magnitude, normalized graph cuts).

For details see our wiki.

First version of Segmentator is out!

It is lacking some documentation and the interface is a bit clunky for now but the code is ready to be used and looks stable enough.

We are already working on some major user interface upgrades for the next version. You will not see as many sliders in the future.