When arriving at the TRACKING page, the image area shows the FA image (A), which highlights areas of high diffusion anisotropy. However, a second image (B) is available for fusion. To see both images, the fusion slider has to be shifted to the right.
In the example below, the FA image is combined with the MD image, which highlights the CSF space. Also shown in the overlay are the VOIs.
Tracking Configuration
The area in the lower right serves for defining the tracking algorithm and its parameters. DTI fiber Tracking algorithms can be divided into DETERMINISTIC and probabilistic methods. PMOD supports the following probabilistic fiber tracking methods: PICO for DTI with pregenerated LUT tables (from the PXMOD module) and Bayesian tracking (on simplified tensors, ball and stick model) directly on DWI images.
The DETERMINISTIC method initiates fiber trajectories from user-defined voxels. For example, in the fiber assignment by continuous tracking algorithm (FACT) the fiber trajectories, also known as “streamlines”, follow the primary eigenvector from voxel to voxel in 3 dimensions. When the fiber trajectory reaches the edge of the voxel, the direction of the trajectory is changed to match the primary eigenvector of the next voxel. Constraints on the maximum turning angle of the streamline between voxels and on the minimum FA within a voxel for propagation of the streamline can be applied to contain the fiber tracks to regions of the brain where the diffusion tensor model realistically represents the white matter pathways[4].
The deterministic streamline fiber tracking is influenced by the noise, subject movement, and distortion from imaging artifacts which produce uncertainty in the orientation of the diffusion ellipsoid[4].
The probabilistic fiber tracking methods compensate the drawback incorporating the expected uncertainty into the tracking algorithm and can be used to produce a connectivity metric for each voxel. Probabilistic fiber tracking techniques tend to disperse trajectories more than deterministic methods and have the potential to delineate a greater portion of a white matter tract. However, the accuracy of these probabilistic methods is still limited by the information contained in the diffusion tensor and the method of constructing the probability density function[4].
For each tracking algorithm three methods are supported and can be selected in the Method list:
1.FACT: Similar to the FACT algorithm proposed by Mori et al [1], this method follows the local fiber orientation in each voxel. No interpolation is used.
2.EULER: Tracking proceeds using a fixed step size along the local fiber orientation [2]. With nearest-neighbor interpolation, this method may be very similar to FACT, except that the step size is fixed, whereas FACT steps extend to the boundary of the next voxel (distance variable depending on the entry and exit points to the voxel).
3.4th order Runge-Kutta method [2]. The step size is fixed, however the eventual direction of the step is determined by taking and averaging a series of partial steps. Interpolation determines how the fiber direction is determined at a given point in space.
The Interpolation setting is not relevant for FACT. For EULER and 4th Runge-Kutta the choices are:
1.NN: Nearest-neighbor interpolation using the local voxel data directly.
2.NC: Probabilistic nearest-neighbor interpolation, similar to the method proposed by Behrens et al [3]. The data is not interpolated, but at each point one of the 8 voxels neighbors is randomly chosen. The probability of choosing a particular voxel is based on how close the point is to the center of that voxel.
3.TRILINEAR: Linear interpolation of the vector components using the 8 neighbors.
Further parameters and options determine how tracking proceeds and when it ends:
Anisotropy |
Track ends if the fractional anisotropy (which is in the range [0,1]) drops below this threshold. |
Step size |
Distance between points on the track. Not used for FACT as it does not apply interpolation. |
Min length |
If checked, tracks shorter than the minimal length specified are discarded. |
Max curvature |
If checked, tracks are terminated in case the curvature exceeds the specified angle within the length specified. If length = 0.0, the curvature is checked at each step. |
Exclusion VOI |
Defines the behavior, when a track enters an EXCLUDE VOI: DISCARD the whole track, or CLIP it at the entry point. |
Colour by |
The track lines can be functionally colored as follows: |
After configuring tracking, please proceed with the Fiber Tracking (3D) button. The tracks are calculated and visualized on the 3D page.
Track Masking
An additional means for restricting the generated tracks is by checking the Mask fibers box and defining a mask. The elements in
have the following functionality:
|
Generate a mask from the FA image by applying the specified threshold and include it in the image list for viewing. |
|
Open the current mask in the VOI editor and convert it into a VOI. The VOI can be edited, and the mask finally updated from the VOI. |
|
Select an external mask file. |
References
1.Mori S, Crain BJ, Chacko VP, van Zijl PC: Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 1999, 45(2):265-269.
2.Basser PJ, Pajevic S, Pierpaoli C, Duda J, Aldroubi A: In vivo fiber tractography using DT-MRI data. Magn Reson Med 2000, 44(4):625-632.
3.Behrens TE, Woolrich MW, Jenkinson M, Johansen-Berg H, Nunes RG, Clare S, Matthews PM, Brady JM, Smith SM: Characterization and propagation of uncertainty in diffusion-weighted MR imaging. Magn Reson Med 2003, 50(5):1077-1088. DOI
4.Mukherjeea P , Bermana JI, Chunga SW, Hessa CP, Henrya RG: Diffusion Tensor MR Imaging and Fiber Tractography: Theoretic Underpinnings. AJNR 2008, 29: 632-641. DOI