Welcome to the home of the "Human-machine integration for vessel segmentation" project.
Please see the following links to learn more about the project:
- About the project.
- List of people involved.
- A list of references related to vessel segmentation, including our publications, which describe the software and results found here.
- Our software for retinal vessel segmentation.
- Downloads of images and our segmentation results.
- The scale-space skeletonization algorithm, implemented and included in mlvessel.
- HTMatLab, an open-source toolbox devoloped within our group for producing HTML output within Matlab, which is distributed with mlvessel.
September 8, 2012 - New version of mlvessel released.
mlvessel 1.4 was released. The new release contains only minor changes, allowing the code to run with the newer versions of Matlab, and making it easier to run for a new user.
January 3, 2009 - Code available at the SVN repository.
The most recent version of our code is now available at our project's SVN repository at Sourceforge.
January 2, 2009 - New website is up!
Our new project website (this website) is now up, powered by Mediawiki. We have basically copied and updated the content from the previous website (http://retina.incubadora.fapesp.br/portal). The project is now hosted by Sourceforge at http://sourceforge.net/projects/retinal/.
July 29, 2008 - Stand-alone executable for the Matlab GUI part of mlvessel.
A Win32 stand-alone version of the mlvessel GUI is now available for download. This version is an executable that does not need an installation of Matlab in order to run, but instead a runtime component, which is included in the package. Please see software for more information and to download.
July 15, 2007 - Draft version of chapter available.
A draft version of a book chapter describing our vessel segmentation approach is now available here. Some details on the use and architecture of the Matlab implementation (mlvessel) are included. The chapter explains our approach and briefly describes some of the theory involved, that is, the two-dimensional continuous wavelet transform and the classifiers that were tested (Bayesian Gaussian mixture model classifier, k-nearest neighbor classifier, and linear minimum squared error classifier).
May 9, 2007 - Trained classifiers and sample images for download.
Trained classifiers and sample images for mlvessel 1.2 are now available for download. The classifiers were trained using images from the STARE and DRIVE databases and work with the graphical user interface of mlvessel 1.2. Also, a pair of sample images and manual segmentations from each database were included for testing. Download the files here (choose mlvessel-optional.zip).