Automatic Fiber Tracking (AutoTrack)

AutoTrack maps named white-matter pathways using DSI Studio’s tractography atlas. It is useful when the goal is to map a specific anatomical bundle consistently across subjects.

Run conventional whole-brain tracking first. AutoTrack should be used only after confirming that the diffusion data, b-table, reconstruction, and general tractography quality are anatomically reasonable.

Streamlines are computational trajectories generated by the tracking algorithm. Their count should not be interpreted as an axon count or a direct measure of biological connection strength.

Basic workflow

1. Open the FIB file

Open the subject .fz file in Step T3: Fiber Tracking.

2. Check whole-brain tractography

Restore the normal tracking settings and generate a representative whole-brain tractogram. Check major pathways and fiber orientations before continuing.

See Whole-Brain Fiber Tracking for the general tracking workflow.

3. Enable AutoTrack

Use [Step T3d: Tracts][Enable AutoTrack] and select the target tract or tract group.

DSI Studio uses the selected tractography atlas and the subject-to-template mapping to recognize streamlines belonging to the requested pathway. AutoTrack entries are hierarchical: choose the parent entry when the whole tract family is intended and a child entry only when a specific subdivision or branch is requested.

Standard named AutoTrack bundles already include atlas-defined anatomical constraints. Do not add extra ROI, ROA, End, Seed, NotEnd, Limiting, or Terminative regions by default. Add an extra region only when the study has a specific anatomical reason, such as isolating a minor branch.

4. Run Fiber Tracking

Click Fiber Tracking to generate the bundle. Inspect the result in multiple views and compare it with expected neuroanatomy.

If a pathway is incomplete or contains implausible branches, first determine whether the problem comes from acquisition/reconstruction quality before adjusting tracking parameters.

Important AutoTrack parameters

Parameter Guidance
Target tract Select the anatomical bundle to map. Use the current atlas entry; select the parent for the whole tract family and a child only for a requested subdivision.
AutoTrack tolerance Controls how far candidate streamlines may deviate from the atlas definition. The current default is derived from the loaded FIB/template. Increase it only when anatomical evidence supports a wider tolerance.
Track/Voxel Ratio Controls the amount of tractography generated. The current default is derived from the FIB data rather than a universal fixed value.
Min Length Removes short fragments. Set it according to the expected pathway and image scale.
Topology-Informed Pruning (TIP) Use TIP on a sufficiently populated, visually coherent bundle. When a bundle reaches roughly 5,000–10,000 or more tracts before pruning, 3–4 iterations are a practical starting point. TIP is not generic whole-brain cleanup.
Check endings AutoTrack can use atlas-defined ending constraints. Keep the default unless the tract definition or study design requires otherwise.

Parameter choices should be kept consistent across subjects in a comparative study.

Troubleshooting AutoTrack

If AutoTrack repeatedly produces no result or poor anatomy, check the upstream data before tuning the tract-recognition settings.

  1. Run SRC quality control.
  2. Verify the image orientation and b-table orientation during reconstruction.
  3. Open the .fz file and confirm that principal fiber directions are anatomically plausible.
  4. Run whole-brain tracking and confirm that major pathways can be reconstructed.
  5. Make sure the selected tractography atlas/template is appropriate for the dataset.

Small or difficult pathways may fail in some subjects when spatial resolution, angular sampling, b-value, SNR, or coverage is insufficient. Relaxing AutoTrack parameters cannot recover anatomical information that is absent from the acquisition.

If an unexpected problem persists, share the relevant SRC/FIB data through the Data Upload link and describe the issue on the DSI Studio forum.

Tractometry

After a tract has been mapped, DSI Studio can quantify its shape and diffusion measurements.

Tract statistics

Use [Tracts][Statistics] to obtain tract-level measurements. Depending on the loaded data, these can include:

For interpretation of diffusion measurements, see How to Interpret dMRI Metrics.

Additional NIfTI measurements such as structural MRI, quantitative MRI, PET, or other maps can be inserted with [Slices][Insert Other Images] and sampled along the tract when they are correctly aligned.

Tract profile

Use [Tracts][Tract Profile] to examine how a measurement changes along a pathway rather than reducing the tract to one mean value.

Profiles can be parameterized by image axes or by fiber orientation. Keep the same profile definition and bandwidth when comparing subjects.

Tract-to-Region (T2R) Connectome

The tract-to-region (T2R) connectome represents connectivity as an n × m matrix, where rows are named white-matter tracts and columns are parcellation regions. This keeps the identity of the connecting white-matter pathway, unlike a conventional region-to-region matrix that summarizes connectivity only between region pairs.

Generate a T2R connectome

1. Map named tract bundles

Use AutoTrack to map the tract bundles included in the analysis. T2R is designed around identifiable pathways, so use named bundles rather than an undifferentiated whole-brain tractogram when the goal is an atlas-based T2R representation.

2. Load a brain parcellation

Use [Step T3a][Atlas] to load the desired parcellation, such as HCP-MMP or another atlas appropriate for the study.

3. Create the matrix

Use [Regions][Tract-to-Region Connectome] to calculate the tract-by-region matrix.

Keep the tract set and parcellation identical across subjects when comparing T2R matrices.

Command-line workflows

For current tractography and intermediate file formats, see DSI Studio File Formats.