HTTPS SSH
Pixel Classification framework for medical images

This code implements the framework presented in [1], which learns a filter bank
on datasets of medical images and then segments them, eventually supported by
the output of two handcrafted algorithms:
- Optimally Oriented Flux (OOF) [3]
- Enhancement Filtering (EF) [4]
A future release of this package will include these two algorithms implemented
in the C++/ITK language. For the moment, to ease the tests, we have pre-computed
the output of those two algorithms for the adopted datasets, and we have put it
at the address [2]. The two packages above can be used to compute the OOF/EF
response for other datasets.

The datasets over which the code was tested are:
- DRIVE [5]
- STARE [6]
- BF2D (which can be found at the project's website [2])
- 2D minimum intensity projections of the VC6 dataset ([7], projections
  are available in this package).

Details about the two components of the framework are given in the respective
subdirectories.

For any question or bug report, please feel free to contact me at:
roberto <dot> rigamonti <at> epfl <dot> ch

If you use this code in your project, please cite our paper ([1]).


[1] R. Rigamonti and V. Lepetit, "Accurate and Efficient Linear Structure
    Segmentation by Leveraging Ad Hoc Features with Learned Filters", MICCAI
    2012
[2] http://cvlab.epfl.ch/~rigamont
[3] http://www.cse.ust.hk/~maxlawwk/bio/bio.htm
[4] http://www.mathworks.com/matlabcentral/fileexchange/24409-hessian-based-frangi-vesselness-filter
[5] http://www.isi.uu.nl/Research/Databases/DRIVE/
[6] http://www.parl.clemson.edu/stare/
[7] http://www.diademchallenge.org/visual_cortical_layer_6_neuron_readme.html