Available tools for 3D segmentation and pixel classification

Please post in this thread links to publicly available software that can help participants segmenting 3D neurites from EM images.

Some options are already publicly available to approach different tasks of the challenge. For example:

Please, feel free to post any other tool that you think may help the other participants.


If your segmentation method produces a 3D affinity graph as output. You can use the following Matlab code by Aleksandar Zlateski to run connected components using a specific threshold value and extract the corresponding 3D objects:

  • ZiConnectedComponents.mexa64: add this file into your Matlab path and you will be able to run connected components on 3D affinities.
    >> R = ZiConnectedComponents(AffinityGraph, Threshold [, MinSize = 0] )
      Affinity - 4D Affinity graph of type float or double
      Threshold - float or double (edges below Threshold are ignored)
      MinSize - float or double or Integer (components smaller than MinSize will be ignored)
  • ZiConnectedComponents.cpp: corresponding C++ source code (GPL license).

For more information about using affinity graphs and image segmentation you can read the following paper, for example.
Check out Fig. 3: http://hebb.mit.edu/people/seung/papers/Turaga%20et%20al_NeuralComp_2010...

If you have any quesiton, just write it on this thread!


There was a small bug on the code. We have just fixed it and uploaded the new files.