Fiber assignment by continuous tracking (FACT) algorithms are a commonly used deterministic post-processing algorithm for magnetic resonance tractography studies.
In these algorithms, axonal fiber bundles are reconstructed - voxel by voxel - following the direction of the main eigenvector. The beginning of this reconstruction process can take place in different ways:
- from a single voxel (seed point)
- from several independent voxels
- from several voxels belonging to the same region of interest (ROI)
The use of a single voxel or several independent voxels is not advisable, because they tend to generate too thin fibers with an approach to the real neural pathway not very reliable. The most reliable approach is the use of several voxels belonging to the same ROI 4,5.
Technically, the fiber-tracking process can be performed both on individual regions and on multiple regions of interests (ROIs). While the fiber-tracking process from single ROI is not very precise (but rapid), that of multiple ROIs, due the possibility of setting constraints in the definition of trajectories, is more accurate. The constraints consist of combining ROIs and using the classic Boolean operators (AND, OR and NOT), refining
thus the fibers delineated by the fiber-tracking process. 5,6
The characteristic of this particular type of algorithm, totally based on data (no interpolation function), is the remarkably fast computational speed. Therefore, although FACT algorithms create some predictable errors, they may in some cases be preferred to more accurate algorithms 2,3.