Source-linked AI summary
Quantitative mapping of the brain's structural connectivity using diffusion MRI tractography: a review
Fan Zhang, Alessandro Daducci, Yong He, Simona Schiavi, Caio Seguin, Robert Smith, Chun-Hung Yeh, Tengda Zhao, Lauren J. O'Donnell
TL;DR
Quantitative dMRI tractography lacks a settled best methodology despite its use for mapping structural connectivity and studying white matter across health and disease. This review synthesizes correction, segmentation, quantification, and application studies, concluding that interpretation should remain cautious because pipeline choices and anatomical biases affect results.
Problem
Quantitative tractography encompasses many methodological choices, while biases and pipeline factors can affect structural-connectivity estimates.
Method
The paper reviews tractography correction, segmentation, and quantification methods, then surveys quantitative tractography applications in health, disease, and neurosurgery.
Results
The review finds considerable methodological and application advances but no consensus about the best quantitative tractography methodology.
Takeaways & Limitations
Researchers should remain cautious when interpreting quantitative tractography results in research and clinical applications.
Takeaways & Limitations
Structural-connectivity matrices are highly consequentially affected by parcellation and other processing-pipeline choices, including streamline assignment and false-positive filtering.
Abstract
from arXiv · showhide
Diffusion magnetic resonance imaging (dMRI) tractography is an advanced imaging technique that enables in vivo mapping of the brain's white matter connections at macro scale. Over the last two decades, the study of brain connectivity using dMRI tractography has played a prominent role in the neuroimaging research landscape. In this paper, we provide a high-level overview of how tractography is used to enable quantitative analysis of the brain's structural connectivity in health and disease. We first provide a review of methodology involved in three main processing steps that are common across most approaches for quantitative analysis of tractography, including methods for tractography correction, segmentation and quantification. For each step, we aim to describe methodological choices, their popularity, and potential pros and cons. We then review studies that have used quantitative tractography approaches to study the brain's white matter, focusing on applications in neurodevelopment, aging, neurological disorders, mental disorders, and neurosurgery. We conclude that, while there have been considerable advancements in methodological technologies and breadth of applications, there nevertheless remains no consensus about the "best" methodology in quantitative analysis of tractography, and researchers should remain cautious when interpreting results in research and clinical applications.
1. Introduction
The review frames dMRI tractography as a way to map white-matter connections in vivo and quantitatively study structural connectivity. It introduces terminology, analysis styles, processing steps, and applications across health and disease.
- dMRI tractography enables in vivo mapping of the brain’s white-matter connections at macro scale.
- The paper begins by listing common terminology and definitions because quantitative tractography has generated a proliferation of terms.
- Quantitative tractography estimates connectivity or microstructural measures for pathways of interest using tract-specific or connectome-based analyses.The categorization is useful but imperfect because some approaches blend both styles.
- Figure 1 illustrates the progression from diffusion-weighted data and streamlines to fiber bundles and a structural connectivity matrix.The matrix represents gray-matter regions by rows and columns, with connection strength quantified here by streamline count.
- The review covers tractography correction, segmentation, and quantification before surveying applications in development, aging, disorders, and neurosurgery.
2. Brief introduction to tractography
Tractography computationally estimates white-matter fiber trajectories from dMRI data through multiple algorithmic approaches. The field includes many software packages, but results remain sensitive to tracking-method choices.
- Tractography refers to computational processes that estimate anatomical white-matter fiber trajectories from dMRI data.
- Available approaches include tensor-based streamline tractography, methods using advanced diffusion models, and machine-learning improvements.
- This review excludes a detailed introduction to tractography algorithms and directs readers to dedicated review papers.
- Many software packages support tractography research, including MRtrix3, FSL, Dipy, TractSeg, and Tracula.
- Tractography is sensitive to the underlying fiber-tracking algorithm, which can affect subsequent quantitative analyses.
3. Improving tractography quality: methods for tractography correction
Tractography-correction methods address biases and errors that can undermine quantitative analysis. The review covers curvature, termination, density, gyral, and false-positive-connection problems, alongside proposed corrections and unresolved trade-offs.
- Correction methods target tractography biases and errors that can compromise fiber-pathway extraction and structural-connectivity matrix construction.
- Curvature overshoot bias: First-order tracking can underestimate curvature in curved bundles, producing erroneous trajectories; higher-order integration directly accounts for curvature but can be technically difficult with crossing-fiber models.
- Termination bias: Diffusion models provide orientation evidence but weak evidence about fiber endpoints, allowing premature termination, fluid entry, or sulcal-bank crossing.
- Connection density biases: Streamline count is not quantitative because reconstructed-streamline numbers do not guarantee correspondence with underlying axon density and depend on nonbiological factors.
- Gyral bias: Gyral bias favors streamline terminations in gyri over sulci, while asymmetric FOD techniques have been shown to alleviate this bias.
- False positive connections: Thresholding removes weak edges to reduce false positives but cannot reliably distinguish true from spurious connections, and filtering methods remain debated.
- False positive connections: Combining anatomical knowledge with data-driven filtering and Group Lasso regularization can improve connectome anatomical accuracy by promoting sparsity among connection groups.
4. Defining regions for quantitation: methods for tractography segmentation
Tractography segmentation defines meaningful white matter pathways for quantitative analysis, using tract-specific or connectome-based strategies. Methods range from manual virtual dissection to automated ROI-based, streamline-labeling, direct-segmentation, cortical-parcellation, and fiber-clustering approaches, with reproducibility and granularity remaining important choices.
- Segmentation goals and analysis styles: Tractography segmentation identifies meaningful white matter pathways for quantifying structural connectivity and supports both tract-specific and connectome-based analyses.It is also used for qualitative applications such as neurosurgical white matter mapping.
- Anatomical tract identification: Manual virtual dissection uses expert-drawn inclusion and exclusion ROIs to select streamlines corresponding to anatomical tracts.Inclusion ROIs define streamline endpoints or passage regions, while exclusion ROIs remove unwanted streamlines.
- Anatomical tract identification: Automated tract identification comprises ROI-based methods, streamline labeling, and direct segmentation, with ROI-based methods reported as the most commonly used.Streamline labeling assigns anatomical labels to individual streamlines, often using geometric similarity, learning models, or fiber clusters.
- Validation and methodological considerations: Validation is difficult because tractography lacks ground truth, so reproducibility and consistency across populations and acquisitions are important evaluation criteria.Manual selection remains the gold standard for benchmark comparisons and algorithm training, while no consensus exists among automated methods.
- Whole-brain tractogram parcellation: Whole-brain parcellation uses cortical-parcellation-based or fiber-clustering methods to organize structural connectivity across the tractogram.Cortical approaches connect gray-matter ROIs, whereas fiber clustering groups streamlines by geometric trajectories.
- Whole-brain tractogram parcellation: Parcellation granularity ranges from tens to millions of parcels, with fine scales suggested for machine learning and statistical analysis and coarse scales for connectome consistency.The cited examples define fine scale as over 2000 parcels and coarse scale as fewer than 200 parcels.
5. Performing quantitative analysis: methods for tractography quantification
Tractography quantification combines connectivity and microstructural measures with strategies that summarize pathways or preserve along-tract variation. The resulting analyses support tract-specific, connectome, and graph-theoretical assessments, but measurements remain sensitive to modeling and processing choices.
- Quantitative measures and extraction: Tractography quantification extracts measures of structural connectivity or white matter microstructure from pathways of interest.The review covers measures, pathway-level extraction, and filtering methods intended to reduce potential biases.
- Quantitative measures and extraction: Tractography-derived measures include streamline counts, tract volume, tract length, and microstructural indices sampled from diffusion or other quantitative imaging models.Common diffusion-model measures include FA, AD, RD, and MD, while advanced models include DKI, Free Water, NODDI, SMT, and AFD.
- Quantitative measures and extraction: Streamline count does not provide a truly quantitative measure of connection strength because dMRI resolution differs from actual axon dimensions.Recent approaches seek to estimate connection density more quantitatively or modulate streamline contributions to aggregate connectivity measures.
- Pathway-level quantification: Pathway measures are commonly reduced to a scalar summary statistic, a strategy required for connectome-based analysis and also used in tract-specific analysis.Other approaches retain the distribution of microstructural measures along the pathway for local analysis.
- Pathway-level quantification: Along-tract analysis maps microstructural measures along fiber pathways to investigate local tissue properties.Related approaches include tractometry, profilometry, and tract-based morphometry; medial-surface methods can also represent data across pathway cross-sections.
- Connectome and graph analysis: Connectivity matrices depend on gray-matter parcellation, streamline-to-parcel assignment, false-positive filtering, and upstream acquisition, modeling, and tractography choices.These factors can affect network topology and connectome sensitivity and specificity.
- Connectome and graph analysis: Graph measures characterize modular organization and global communication, but different communication models encode different assumptions about how signals traverse structural connectivity.Shortest-path measures require regions to possess global connectivity knowledge, while the biological accuracy of alternative measures remains an open question.
6. Applications of quantitative tractography analysis
Quantitative tractography is applied across development, aging, neurological and mental disorders, and neurosurgery to characterize white-matter maturation, degeneration, dysconnectivity, and surgical anatomy. Findings span tract-specific diffusion changes and network-level topology, while tractography also provides clinically useful spatial guidance.
- Developmental tractography: Prenatal tractography identifies neuronal migration pathways and emerging major tracts, with all major structural pathways identifiable by term.Fornix and cingulum parts appear by 13 weeks, while part of the corpus callosum appears by 15 weeks.
- Developmental tractography: Postnatal studies consistently report age-related increases in FA and MD across widespread fiber tracts, alongside nonlinear developmental trajectories.These studies span infancy through adolescence and use both cross-sectional and longitudinal designs.
- Aging and lifespan tractography: Across the lifespan, aging is associated with decreased FA, increased MD, and reduced tract volume, with prefrontal association tracts especially vulnerable.Corpus-callosum FA decline may be localized to the genu rather than the body or splenium.
- Aging and lifespan tractography: Structural network layouts remain largely preserved into old age, while network efficiency and small-world measures follow inverted U-shaped trajectories peaking at 30 years.Regional maturation also differs, with default-mode-network regions displaying later peak maturation.
- Neurological and mental disorders: In multiple sclerosis, lesions disrupt network integration, long-range communication, rich-club connectivity, and cognitive performance, while network segregation may increase after the first clinical event.Global-efficiency changes are significantly correlated with Expanded Disability Status Scale scores.
- Neurological and mental disorders: In schizophrenia, tract-specific frontotemporal and callosal abnormalities coincide with altered global topology, impaired integration, enhanced segregation, and associations with symptoms and cognition.Reduced network efficiency appears in chronic and first-episode patients, unaffected siblings, and high-risk infants compared with healthy controls.
- Tractography in neurosurgery: In neurosurgery, tractography localizes displaced white-matter pathways and supports lesion targeting, function preservation, and network-guided oncological disconnection surgery.Combined with navigated transcranial magnetic stimulation, it aids speech preservation near the superior longitudinal fasciculus.
7. Discussion and conclusion
The review finds substantial advances in quantitative tractography methods and applications, but no consensus on a best methodology. Anatomical inaccuracies and the indirect biological meaning of tractography outputs require caution in research and clinical interpretation.
- The review covered methods for the main processing steps of quantitative tractography and studies applying them to brain white matter.
- Biological interpretation requires extreme caution because streamlines are simulated entities and diffusion metrics infer local properties rather than directly measuring tissue properties.
- Anatomical accuracy remains an ongoing challenge because tractography can produce both false-positive and false-negative tracking results.
- Quantitative tractography has advanced methodologically and broadened its applications across the lifespan and in health and disease.
- No consensus currently identifies a single best methodology for quantitative tractography analysis.