The TractoR (Tractography with R) project includes R packages for reading, writing and visualising magnetic resonance images stored in Analyze, NIfTI and DICOM file formats (DICOM support is read only). It also contains functions specifically designed for working with diffusion MRI and tractography, including a standard implementation of the neighbourhood tractography approach to white matter tract segmentation. A shell script is also provided to run experiments with TractoR without interacting with R. Using TractoR you can easily
- Convert DICOM files from your MR scanner to Analyze/NIfTI format (see ScriptGroups#File_reading_and_conversion).
- Preprocess diffusion MR data to calculate tensor metrics including fractional anisotropy (FA), mean diffusivity (MD), and principal directions (see Preprocessing).
- Run probabilistic tractography using single seed points or one or more masks (see ScriptGroups#Conventional_tractography).
- Segment specific tracts in groups automatically using neighbourhood tractography (see PNTTutorial and HNTTutorial).
- Remove false positive tracts using a model of tract shape variability (see PNTTutorial#Visualising_and_interpreting_results).
- Create graphics to visualise image slices or maximum-intensity projections (see ScriptGroups#Visualisation).
Please note that TractoR is research software and has not been approved for clinical use. The software is provided in the hope that it will be useful, but comes with no warranty whatsoever.
General queries or problems may be sent to the users' mailing list, while bugs and other specific issues may be reported using the Issues tab above. Please describe any problem as fully as possible.
News
(2009-11-10) Version 1.3.0 of TractoR is now available. This version adds a new method for rejecting false positive streamlines using a PNT model, which substantially improves tract segmentation; and includes a set of tests to ensure that TractoR is functioning properly on your system. There are also various minor enhancements, as usual, and these are documented in the Changelog.
(2009-08-21) TractoR version 1.2.0 has just been released. It includes seven new reference tracts for both HNT and PNT, as well as DICOM performance improvements and various other minor enhancements. As ever, details are in the Changelog.
(2009-07-03) Documentation aimed at existing R users, which describes the R packages provided by TractoR, as well as how to write new TractoR-compatible R scripts for use with the command line tractor shell script, is now available.
(2009-06-15) TractoR version 1.1.0 has been released. This version contains a number of improvements to various aspects of the software, including support for a more general variant of the standard PNT tract model. Full details are in the Changelog.
(2009-05-29) Updates to the documentation, particularly the HNT and PNT tutorials, have been made to reflect the current state of the software. Additional information about reference tracts has also been added.
(2009-05-14) The "tractor.base" R package is now available on the Comprehensive R Archive Network (CRAN). It can be found here, or through your favourite CRAN mirror (once updated). This allows R developers to use the TractoR functions for reading, writing, manipulating and visualising magnetic resonance images in their own code. A short introduction to TractoR for R users will be made available here soon.
(2009-02-18) The tractor-users mailing list can now be joined by anyone. This list/group is for any discussion relating to the software. Announcements of new versions of TractoR are also made to the list.
(2009-02-13) TractoR version 1.0.0 has been released. New features in this version include "unsupervised" neighbourhood tractography (NT), a set of standard reference tracts for both heuristic and probabilistic NT, and more flexible conversion of DICOM-format files. As usual, the Changelog has full details on the changes since the last beta version. Although this is no longer a beta version, please remember that TractoR is research software and comes with no warranty. Dropping the "beta" label reflects mainly that the structure of the package has stabilised somewhat.
