ScholarGate
Get started
Process / pipelineNeuroimagingVoxel-wise spatial analysisPipeline

Tract-Based Spatial Statistics

Also known as: TBSS, white matter skeleton analysis

OriginatorStephen M. SmithYear2006Sources2Related methods7

Tract-Based Spatial Statistics (TBSS) is a voxel-wise analysis method for detecting group differences in white matter microstructure from diffusion MRI data. Published by Stephen M. Smith and colleagues in 2006, TBSS addresses registration and multiple comparison problems inherent in voxel-wise analysis by projecting individual FA maps onto a white matter skeleton derived from a population template.

Key highlights

  • Reduces alignment uncertainty by focusing on the white matter skeleton, the most anatomically consistent region
  • Improves statistical power by increasing the proportion of voxels that are truly homologous across subjects
  • Addresses multiple comparison problem more efficiently than standard voxel-wise analysis
  • Easily accommodates multiple metrics (FA, MD, AD, RD) in a unified framework
  • Robust to subtle differences in brain size and shape across subjects

Intuition

This section is available to Pro members. Upgrade to Pro

How it works

This section is available to Pro members. Upgrade to Pro

When to use it

TBSS is ideal for group studies comparing white matter integrity between populations, when sample sizes are moderate to large (20+ per group), and when interest is in FA or other diffusion metrics. TBSS is less suitable for single-case clinical assessment or when the primary focus is on specific named tracts (alternative: ROI-based approaches). Avoid TBSS if subject motion is severe or structural pathology is extensive.

Strengths & limitations

Strengths
  • Reduces alignment uncertainty by focusing on the white matter skeleton, the most anatomically consistent region
  • Improves statistical power by increasing the proportion of voxels that are truly homologous across subjects
  • Addresses multiple comparison problem more efficiently than standard voxel-wise analysis
  • Easily accommodates multiple metrics (FA, MD, AD, RD) in a unified framework
  • Robust to subtle differences in brain size and shape across subjects
Limitations
  • Skeleton projection may miss group differences at the periphery of white matter bundles
  • Still requires successful registration across all subjects; pathological brains may not register well to template
  • Less sensitive to crossing fibers and regions with complex architecture (e.g., internal capsule)
  • Interpretation limited to scalar metrics (FA, MD); does not model fiber direction or connectivity

Common pitfalls

This section is available to Pro members. Upgrade to Pro

Applications

This section is available to Pro members. Upgrade to Pro

Frequently asked

What is the white matter skeleton, and why is it important?

The skeleton is the medial core of white matter bundles, representing voxels present in most subjects after registration. It is the most anatomically consistent region across subjects, reducing registration error and improving statistical power. The skeleton approach is TBSS's key innovation.

Which diffusion metric should I analyze with TBSS?

Fractional anisotropy (FA) is most commonly used and most robust. Mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) can also be analyzed. Choose based on your hypothesis: FA captures overall white matter integrity, RD suggests demyelination, and AD reflects axonal damage.

How do I choose the skeleton threshold?

Threshold is usually set to include voxels with FA > 0.2 in the population mean FA map. This threshold is a compromise: it includes most white matter while excluding gray matter and CSF. Sensitivity analysis around this value can test robustness.

Can TBSS handle patient groups with brain lesions?

TBSS registration can be problematic if pathology causes large distortions (e.g., tumors, infarcts). Consider masking lesions during registration or using lesion-avoiding registration tools. For severe pathology, ROI-based or tract-specific approaches may be more reliable.

Sources

  1. 1.
    Smith, S. M., Jenkinson, M., Johansen-Berg, H., et al. (2006). Tract-based spatial statistics: voxelwise analysis of multi-subject diffusion data. NeuroImage, 31(4), 1487–1505.
    DOI 10.1016/j.neuroimage.2006.02.024
  2. 2.
    Winkler, A. M., Ridgway, G. R., Webster, M. A., Smith, S. M., & Nichols, T. E. (2014). Permutation inference for the general linear model. NeuroImage, 92, 381–397.
    DOI 10.1016/j.neuroimage.2014.01.060

You have read it. What now?

Ask about this methodSimilar methods

Cite this page

ScholarGate. (2026, June 3). Tract-Based Spatial Statistics. ScholarGate. https://scholargate.app/neuroimaging/tract-based-spatial-statistics

Tract-Based Spatial Statistics | ScholarGate