Pith. sign in

REVIEW 3 major objections 6 minor 40 references

A conditional generative model that samples all white-matter streamlines in parallel achieves 2.1× higher precision than the next-best tractography method, and an order of magnitude more under noise.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-03 21:51 UTC pith:CH4CNQZP

load-bearing objection GenTract offers a genuinely new global-generative twist on tractography with large reported precision gains, but the design isn't yet clean enough to separate conditioning from bundle-prior memorization. the 3 major comments →

arxiv 2511.13183 v2 pith:CH4CNQZP submitted 2025-11-17 cs.CV

GenTract: Generative Global Tractography

Alec Sargood , Lemuel Puglisi , Elinor Thompson , Mirco Musolesi , Daniel C. Alexander This is my paper
classification cs.CV
keywords tractographygenerative modeldiffusion modelflow matchingwhite matterdMRIstreamline generationglobal tractography
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

GenTract reframes whole-brain tractography as a generative task: instead of tracing streamlines step-by-step through local fiber orientation estimates, it learns a direct mapping from the entire diffusion-weighted volume to complete, anatomically plausible streamlines, generating all coordinates simultaneously. The paper claims this is the first generative model for global tractography, and reports precision 2.1× higher than the next-best deep learning baseline on high-quality data, with the gap widening to an order of magnitude on low-resolution, noisy scans, at inference times comparable to other learning-based methods. A sympathetic reader would care because it suggests global tractography—traditionally computationally prohibitive and error-prone—can become practical on clinical-grade data.

Core claim

GenTract learns to sample streamlines directly from a conditional distribution over whole-brain fiber orientation data. A Transformer, conditioned on a learned embedding of the spherical-harmonic coefficients of the dMRI signal, generates the 3D coordinates of all points on a streamline in parallel, producing complete tractograms without any seeding mask. On an independent atlas-based precision metric, it reaches 61.95% versus 28.93% for the next-best baseline, and under low-resolution plus noise it retains 15.73% versus 1.12%, while recovering fewer of the 51 canonical bundles (36.6 vs 48.2). The authors attribute the precision gain to global conditioning and parallel generation, which avoi

What carries the argument

The central object is a conditional generative model: a Transformer that takes a learned embedding of the entire dMRI volume and a noise vector, and outputs the coordinates of all streamline points simultaneously. The conditioning tensor is built from per-coefficient autoencoders over spherical-harmonic fiber orientation volumes, fused by a shared class-conditioned encoder, and the Transformer uses self-attention over points along a streamline and cross-attention over the global context. This replaces the stepwise error propagation of local tracking and the energy optimization of classical global tractography with a single parallel sampling process, and removes the need for a seeding mask.

Load-bearing premise

The training target is a filtered tractogram restricted to 24 atlas bundles, so the model's high precision is measured against the same type of filtering and may exaggerate biological accuracy if that filtering omits genuine white-matter pathways.

What would settle it

Generate a tractogram on real clinical low-field data and have an independent anatomical expert mark false positives; if GenTract's precision falls to the same level as stepwise methods when scored against a 100-bundle atlas or expert consensus, the reported advantage is an artifact of the restricted training/filtering distribution.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Whole-brain tractograms can be generated in one forward pass, making global tractography fast enough for clinical use.
  • Precision on noisy, low-resolution scans is an order of magnitude above the best existing baseline, suggesting connectivity mapping is feasible on routine clinical dMRI.
  • Eliminating the seeding mask removes a major source of operator-dependent variability, improving reproducibility across sites and operators.
  • Because both diffusion and flow-matching objectives work, the architecture can absorb advances in generative modeling without requiring a new tractography-specific design.
  • The precision-versus-bundle-recall trade-off points to the training distribution as the limiting factor, not the generative formulation itself.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The reported precision gain likely reflects a strong prior over anatomically plausible bundles learned from the filtered 24-bundle training set; training on a broader, less filtered distribution would test whether the gain survives.
  • The method's resilience to synthetic noise and resolution loss suggests it captures geometric priors rather than local signal patterns, which may transfer to other dMRI acquisition artifacts such as subject motion.
  • The same global-conditioning design could be adapted to other streamline-like structure inference problems, such as vascular or neuronal reconstruction.
  • A decisive validation would be a comparison on real low-field clinical scans, since synthetic degradation may not fully reproduce the characteristics of true clinical noise and blurring.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The manuscript presents GenTract, a conditional generative model for whole-brain tractography. Diffusion-weighted MRI is represented as per-coefficient SH volumes, compressed by per-coefficient VAEs and a class-conditioned encoder into a latent tensor z; a transformer then generates streamline coordinates in parallel, trained with either a diffusion or flow-matching objective. The authors evaluate on HCP test subjects and an external TractoInferno dataset under original, noisy, and low-resolution+noisy conditions, comparing against classical local (iFOD2, SD Stream), RL-based (TractOracle), deep generative local (DDTracking), and classical global (tckglobal) baselines. They report that GenTract achieves substantially higher precision (BundleSeg and TO-Net) at the cost of lower bundle recall, and shows less degradation in low-quality settings.

Significance. The paper introduces a genuinely new paradigm—treating tractography as a single conditional generation step rather than local sequential tracking—and the empirical evaluation is more careful than typical in this area: the main comparison uses two independent evaluation tools (BundleSeg and TO-Net) that were not used to construct training targets, and it includes an external multi-site dataset. If the conditional-generation mechanism is shown to be genuinely driven by the dMRI conditioning, the work would be an important advance in robustness for tractography. The reported precision margins (2.1x over TractOracle on original HCP, order-of-magnitude under LR+noise) are striking. However, the missing conditioning ablation and the constrained training distribution are significant gaps that must be addressed before the central claims can be accepted.

major comments (3)
  1. [§4.2 and §5.3] The manuscript does not demonstrate that the anatomical conditioning tensor z actually drives the generated streamlines. All experiments condition on z; no ablation with zeroed, shuffled, or subject-mismatched z is reported. Because the training set itself is a strongly constrained distribution (PyAFQ-filtered 24-bundle tractograms), a model that ignores z and merely samples from the training prior could still score high BS % P in Table 2, since BundleSeg rewards bundle-like, atlas-aligned streamlines. This is load-bearing for the central claim of a 'conditional generative model' and 'direct mapping from dMRI to streamlines.' Please add an unconditional or mis-conditioned baseline (e.g., z=0, z from a different subject) and report the same metrics; without it, the reader cannot distinguish true conditioning from prior memorization.
  2. [§5.1, Table 2, §6] The abstract and Introduction describe the output as 'complete, anatomically plausible streamlines' and 'whole-brain tractograms,' but the training targets are PyAFQ-filtered tractograms restricted to 24 atlas bundles, and Table 2 shows GenTract recovers only 36.6/51 bundles on the BundleSeg atlas versus 48.2 for TractOracle. The Discussion (Section 6) correctly acknowledges a 'constrained bundle distribution,' but this contradicts the whole-brain framing. Please revise the claims to reflect the limited bundle coverage, or retrain/evaluate on an unfiltered or more complete training target.
  3. [§4.1] The conditioning encoder is a central component, but its output z is never validated. No reconstruction quality, latent-space analysis, or evidence that z encodes subject-specific anatomical information is provided. Without such validation, the reader cannot assess whether the global conditioning is meaningful or whether the VAE latents are effectively noise. At minimum, please report reconstruction metrics and show that z varies across subjects and correlates with known anatomy (e.g., bundle presence).
minor comments (6)
  1. [Abstract] The phrase '1.8x and 2.1x higher than the next-best methods, DDTracking and TractOracle, respectively' is inconsistent with Table 2: DDTracking's BS % P is 0.49 against GenTract's 61.95, a factor of ~126. Please correct this numerical error.
  2. [§5.2] The definition of precision as the percentage of streamlines retained by a filtering tool is acknowledged as a proxy, but the term 'True Positive' is somewhat misleading when no ground-truth tract exists; suggest renaming to 'retained' to avoid implying a biological ground truth.
  3. [§5.1] The text states that 'all streamline coordinates are min-max scaled to [-1,1] using statistics computed from the training set,' but it is not clear whether the same affine transformation is applied independently per subject; please clarify.
  4. [General] The paper repeatedly refers to the Supplementary Material for implementation details and statistical tests, but no supplementary material is included in the arXiv submission. Please ensure the supplementary is provided.
  5. [Table 2] DDTracking's bundle count (9.30) is much lower than other methods; since DDTracking is a deep generative local method, it may not be designed for whole-brain tractography. A sentence explaining this low recall would help.
  6. [§5.3] Model selection is performed on AFQ % P, which is the same filter used for training. Using an independent metric for configuration selection would strengthen the claim that the chosen architecture is not overfit to the training filter.

Circularity Check

1 steps flagged

Internal model selection is circular via the PyAFQ-only metric; the headline BS/TO-Net precision remains externally evaluated.

specific steps
  1. self definitional [Section 5.3, Table 1 (Internal Model Comparison Study); cf. Section 5.1]
    "For the internal model comparisons, we evaluate tractograms using the PyAFQ filtering mechanism [16]. This is the same filtering mechanism used to filter the supervised training data, and we use this tool to calculate both precision (AFQ % P) and number of recovered bundles (AFQ Bundles, up to 24)."

    Section 5.1 constructs the supervised target with the PyAFQ pipeline, so every training streamline is by definition a PyAFQ-retained streamline. The internal selection metric (AFQ % P) is the retention rate of that same filter. A model that memorizes the 24-bundle PyAFQ training distribution will score near 100% AFQ % P on in-distribution test subjects, making the metric a measure of reproducing the training filter rather than of dMRI-conditioned anatomical inference. The architecture choice in Table 1 is therefore selected to match the PyAFQ prior, and this selection propagates to all subsequent experiments.

full rationale

GenTract's central SOTA precision claim is not reduced to its training input by equation: Section 5.4 deliberately evaluates BS % P and TO-Net % P with tools not used to build the training targets (BundleSeg and TO-Net), and the Discussion acknowledges that no biological ground truth exists and that proxy evaluation tools can introduce their own biases. I find no load-bearing self-citation: the only author-overlapping reference ([4], a handbook chapter) is used for background, not to justify GenTract's design choices or uniqueness. However, there is one genuine circular step in the internal selection: Section 5.3 ranks architectures by AFQ % P, which is exactly the PyAFQ filtering mechanism that generated the supervised training distribution in Section 5.1. On in-distribution test subjects, reproducing the PyAFQ-filtered 24-bundle prior yields high AFQ % P by construction, so the selected Diffusion M=8 n=256 configuration is partly chosen to fit the training filter rather than by independent anatomical evidence. The headline results use independent metrics, so the circularity is partial and limited to model selection; the residual risk that BundleSeg and TO-Net reward the same canonical-bundle geometry the model was trained on is a construct-overlap limitation rather than a strict definitional reduction.

Axiom & Free-Parameter Ledger

3 free parameters · 5 axioms · 0 invented entities

No new physical entities; the model introduces a learned conditioning tensor z, which is a data representation, not a postulated entity. The main hidden input is the filtered training target.

free parameters (3)
  • Model configuration (M=8 transformer layers, n=256 embedding, 10 DDIM steps) = M=8, n=256, 10 steps
    Selected on internal AFQ % P (Table 1, Figure 3), which is the same PyAFQ filter used to generate training targets; hand-chosen by validation.
  • Synthetic degradation parameters (Rician noise σ=0.005, 3mm isotropic downsampling) = σ=0.005, 3mm
    Taken from prior studies (refs 10, 39) and applied to create robustness test sets; not a derivation parameter but affects the evaluation.
  • SH Lmax=6 (m=28 coefficients) = 28 coefficients
    Standard fODF representation choice; determines input dimensionality.
axioms (5)
  • standard math The SH expansion f(θ,ϕ) ≈ Σ α_lk Y_l^k (Eq 1) adequately represents the fiber orientation distribution
    Foundation of input representation; standard in tractography.
  • standard math Diffusion and flow matching objectives (Eqs 2,3) learn the conditional streamline distribution
    Theoretical grounding of the generative training; standard ML.
  • domain assumption PyAFQ-filtered tractograms restricted to 24 bundles are a valid supervision target for white-matter anatomy
    Load-bearing: the model is trained solely on this filtered distribution (Section 5.1), so any bias in PyAFQ filtering becomes a bias in GenTract's output.
  • domain assumption BundleSeg and TO-Net are valid independent proxies for streamline plausibility
    The SOTA evaluation relies on these tools to define precision; they may have their own biases.
  • ad hoc to paper The VAE latent z and fused tensor z (Section 4.1) preserve dMRI information sufficient for tractography
    Architectural assumption that compression does not discard fiber-relevant information.

pith-pipeline@v1.3.0-alltime-deepseek · 12287 in / 10627 out tokens · 94209 ms · 2026-08-03T21:51:25.942232+00:00 · methodology

0 comments
read the original abstract

Tractography is the process of inferring the trajectories of white-matter pathways in the brain from diffusion magnetic resonance imaging (dMRI). Local tractography methods, which construct streamlines by following local fiber orientation estimates stepwise through an image, are prone to error accumulation and high false positive rates, particularly on noisy or low-resolution data. In contrast, global methods, which attempt to optimize a collection of streamlines to maximize compatibility with underlying fiber orientation estimates, are computationally expensive. To address these challenges, we introduce GenTract, the first generative model for global tractography. We frame tractography as a generative task, learning a direct mapping from dMRI to complete, anatomically plausible streamlines. We compare both diffusion-based and flow matching paradigms and evaluate GenTract's performance against state-of-the-art baselines. Notably, GenTract achieves precision 1.8x and 2.1x higher than the next-best methods, DDTracking and TractOracle, respectively. This advantage becomes even more pronounced in challenging low-resolution and noisy settings, where it outperforms the closest competitor by a factor of 3.5. By producing tractograms with high precision on research-grade data while also maintaining reliability on imperfect, lower-resolution data, GenTract represents a promising solution for global tractography.

Figures

Figures reproduced from arXiv: 2511.13183 by Alec Sargood, Daniel C. Alexander, Elinor Thompson, Lemuel Puglisi, Mirco Musolesi.

Figure 1
Figure 1. Figure 1: Overview of our proposed GenTract methodology. We [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the GenTract framework. A: VAEs encode fODF coefficients into latent representations. B(1) and B(2): The training protocol for Diffusion and FM models respectively, using the learned z (i) as input. Losses are back-propagated through both the generative model and the class-conditioned encoder E c . C: The inference process, where streamlines are generated by sampling from Gaussian noise conditi… view at source ↗
Figure 3
Figure 3. Figure 3: AFQ streamlines retained (AFQ % P) vs. Computational [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative result showing SLFR segmented by Bundle [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Computational time comparison for all methods. Bars [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

40 extracted references · 3 linked inside Pith

  1. [1]

    Nvidia hopper h100 gpu: Scaling perfor- mance.IEEE Micro, 43(3), 2023

    Jack Choquette. Nvidia hopper h100 gpu: Scaling perfor- mance.IEEE Micro, 43(3), 2023. 4

    2023

  2. [2]

    Global tractogra- phy of multi-shell diffusion-weighted imaging data using a multi-tissue model.NeuroImage, 123, 2015

    Daan Christiaens, Marco Reisert, Thijs Dhollander, Stefan Sunaert, Paul Suetens, and Frederik Maes. Global tractogra- phy of multi-shell diffusion-weighted imaging data using a multi-tissue model.NeuroImage, 123, 2015. 2, 3, 6

    2015

  3. [3]

    Dipy, a library for the analysis of diffusion MRI data.Frontiers in Neuroinformatics, 8, 2014

    Dipy Contributors, Eleftherios Garyfallidis, Matthew Brett, Bagrat Bagrat Amirbekian, Ariel Rokem, Stefan van der Walt, Maxime Descoteaux, and Ian Nimmo-Smith. Dipy, a library for the analysis of diffusion MRI data.Frontiers in Neuroinformatics, 8, 2014. 5

    2014

  4. [4]

    Global tractogra- phy

    Alessandro Daducci, Simona Schiavi, Daan Christiaens, Robert Smith, and Daniel C Alexander. Global tractogra- phy. InHandbook of Diffusion MR Tractography. Elsevier,

  5. [5]

    Deterministic and probabilistic trac- tography based on complex fibre orientation distributions

    Maxime Descoteaux, Rachid Deriche, Thomas R Knosche, and Alfred Anwander. Deterministic and probabilistic trac- tography based on complex fibre orientation distributions. IEEE Transactions on Medical Imaging, 28(2), 2008. 1, 2, 6

    2008

  6. [6]

    The early development of brain white matter: a review of imaging studies in fetuses, newborns and infants

    Jessica Dubois, Ghislaine Dehaene-Lambertz, Sofya Ku- likova, Cyril Poupon, Petra S H ¨uppi, and Lucie Hertz- Pannier. The early development of brain white matter: a review of imaging studies in fetuses, newborns and infants. Neuroscience, 276, 2014. 1

    2014

  7. [7]

    White matter in learning, cognition and psychiatric disorders.Trends in Neurosciences, 31(7), 2008

    R Douglas Fields. White matter in learning, cognition and psychiatric disorders.Trends in Neurosciences, 31(7), 2008. 1

    2008

  8. [8]

    A novel global tractography algorithm based on an adaptive spin glass model

    Pierre Fillard, Cyril Poupon, and Jean-Franc ¸ois Mangin. A novel global tractography algorithm based on an adaptive spin glass model. InProceedings of the 12th International Conference on Medical Image Computing and Computer- Assisted Intervention. Springer, 2009. 2

    2009

  9. [9]

    White matter and cognition: making the connection.Journal of Neuro- physiology, 116(5), 2016

    Christopher M Filley and R Douglas Fields. White matter and cognition: making the connection.Journal of Neuro- physiology, 116(5), 2016. 1

    2016

  10. [10]

    Diffu- sion Tensor MRI and Spherical-Deconvolution-Based Trac- tography on an Ultra-Low Field Portable MRI System.arXiv preprint arXiv:2506.04473, 2025

    James Gholam, Phil Schmid, Joshua Ametepe, Alix Plum- ley, Leandro Beltrachini, Francesco Padormo, Rui Teixeira, Rafael OHalloran, Kaloian Petkov, Klaus Engel, et al. Diffu- sion Tensor MRI and Spherical-Deconvolution-Based Trac- tography on an Ultra-Low Field Portable MRI System.arXiv preprint arXiv:2506.04473, 2025. 5

    Pith/arXiv arXiv 2025

  11. [11]

    Maisi: Medi- cal ai for synthetic imaging

    Pengfei Guo, Can Zhao, Dong Yang, Ziyue Xu, Vish- wesh Nath, Yucheng Tang, Benjamin Simon, Mason Belue, Stephanie Harmon, Baris Turkbey, et al. Maisi: Medi- cal ai for synthetic imaging. InProceedings of the 2025 IEEE/CVF Winter Conference on Applications of Computer Vision. IEEE, 2025. 3, 4

    2025

  12. [12]

    Denoising diffu- sion probabilistic models.Advances in Neural Information Processing Systems, 33, 2020

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models.Advances in Neural Information Processing Systems, 33, 2020. 3

    2020

  13. [13]

    A bayesian framework for global tractography

    Saad Jbabdi, Mark W Woolrich, Jesper LR Andersson, and TEJ Behrens. A bayesian framework for global tractography. NeuroImage, 37(1), 2007. 2

    2007

  14. [14]

    Diffusion MRI Fiber Tractography of the Brain.NMR in Biomedicine, 32(4), 2019

    Ben Jeurissen, Maxime Descoteaux, Susumu Mori, and Alexander Leemans. Diffusion MRI Fiber Tractography of the Brain.NMR in Biomedicine, 32(4), 2019. 2

    2019

  15. [15]

    Advancements in diffusion MRI tractography for neuro- surgery.Investigative Radiology, 59(1), 2024

    Koji Kamagata, Christina Andica, Wataru Uchida, Kaito Takabayashi, Yuya Saito, Matthew Lukies, Akifumi Hagi- wara, Shohei Fujita, Toshiaki Akashi, Akihiko Wada, et al. Advancements in diffusion MRI tractography for neuro- surgery.Investigative Radiology, 59(1), 2024. 1

    2024

  16. [16]

    Evaluating the reliability of human brain white matter trac- tometry.Aperture Neuro, 1(1), 2021

    John Kruper, Jason D Yeatman, Adam Richie-Halford, David Bloom, Mareike Grotheer, Sendy Caffarra, Gregory Kiar, Iliana I Karipidis, Ethan Roy, Bramsh Q Chandio, et al. Evaluating the reliability of human brain white matter trac- tometry.Aperture Neuro, 1(1), 2021. 2, 5

    2021

  17. [17]

    Filtering in tractography using autoen- coders (FINTA).Medical Image Analysis, 72, 2021

    Jon Haitz Legarreta, Laurent Petit, Franc ¸ois Rheault, Guil- laume Theaud, Carl Lemaire, Maxime Descoteaux, and Pierre-Marc Jodoin. Filtering in tractography using autoen- coders (FINTA).Medical Image Analysis, 72, 2021. 5

    2021

  18. [18]

    DDTracking: A Deep Generative Framework for Diffusion MRI Tractography with Streamline Local-Global Spatiotemporal Modeling.arXiv preprint arXiv:2508.04568, 2025

    Yijie Li, Wei Zhang, Xi Zhu, Ye Wu, Yogesh Rathi, Lau- ren J O’Donnell, and Fan Zhang. DDTracking: A Deep Generative Framework for Diffusion MRI Tractography with Streamline Local-Global Spatiotemporal Modeling.arXiv preprint arXiv:2508.04568, 2025. 1, 2, 6

    arXiv 2025

  19. [19]

    Flow Matching for Generative Modeling.arXiv preprint arXiv:2210.02747, 2022

    Yaron Lipman, Ricky TQ Chen, Heli Ben-Hamu, Maxim- ilian Nickel, and Matt Le. Flow Matching for Generative Modeling.arXiv preprint arXiv:2210.02747, 2022. 3

    Pith/arXiv arXiv 2022

  20. [20]

    Aging of cerebral white matter.Ageing Research Reviews, 34, 2017

    Huan Liu, Yuanyuan Yang, Yuguo Xia, Wen Zhu, Rehana K Leak, Zhishuo Wei, Jianyi Wang, and Xiaoming Hu. Aging of cerebral white matter.Ageing Research Reviews, 34, 2017. 1

    2017

  21. [21]

    The challenge of mapping the human con- nectome based on diffusion tractography.Nature Communi- cations, 8(1), 2017

    Klaus H Maier-Hein, Peter F Neher, Jean-Christophe Houde, Marc-Alexandre Cˆot´e, Eleftherios Garyfallidis, Jidan Zhong, Maxime Chamberland, Fang-Cheng Yeh, Ying-Chia Lin, Qing Ji, et al. The challenge of mapping the human con- nectome based on diffusion tractography.Nature Communi- cations, 8(1), 2017. 1, 2

    2017

  22. [22]

    A framework based on spin glass models for the inference of anatomical connec- tivity from diffusion-weighted MR data–a technical review

    Jean-Franc ¸ois Mangin, Cyril Poupon, Yann Cointepas, De- nis Rivi`ere, Dimitri Papadopoulos-Orfanos, Chris A Clark, Jean R ´egis, and Denis Lrke Bihan. A framework based on spin glass models for the inference of anatomical connec- tivity from diffusion-weighted MR data–a technical review. NMR in Biomedicine: An International Journal Devoted to the Develo...

    2002

  23. [23]

    Toward global tractography.NeuroImage, 80, 2013

    Jean-Franc ¸ois Mangin, Pierre Fillard, Yann Cointepas, Denis Le Bihan, Vincent Frouin, and Cyril Poupon. Toward global tractography.NeuroImage, 80, 2013. 1

    2013

  24. [24]

    Fiber tractography using machine learning.NeuroImage, 158,

    Peter F Neher, Marc-Alexandre C ˆot´e, Jean-Christophe Houde, Maxime Descoteaux, and Klaus H Maier-Hein. Fiber tractography using machine learning.NeuroImage, 158,

  25. [25]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. Scalable diffusion models with transformers. InProceedings of the IEEE/CVF Interna- tional Conference on Computer Vision, 2023. 4

    2023

  26. [26]

    Generative AI for medical imaging: extending the MONAI framework.arXiv preprint arXiv:2307.15208, 2023

    Walter HL Pinaya, Mark S Graham, Eric Kerfoot, Petru- Daniel Tudosiu, Jessica Dafflon, Virginia Fernandez, Pedro Sanchez, Julia Wolleb, Pedro F Da Costa, Ashay Patel, et al. Generative AI for medical imaging: extending the MONAI framework.arXiv preprint arXiv:2307.15208, 2023. 4

    Pith/arXiv arXiv 2023

  27. [27]

    TractoInferno-A large-scale, open-source, multi-site database for machine learning dMRI tractography

    Philippe Poulin, Guillaume Theaud, Francois Rheault, Eti- enne St-Onge, Arnaud Bore, Emmanuelle Renauld, Louis de Beaumont, Samuel Guay, Pierre-Marc Jodoin, and Maxime Descoteaux. TractoInferno-A large-scale, open-source, multi-site database for machine learning dMRI tractography. Scientific Data, 9(1), 2022. 5

    2022

  28. [28]

    Global fiber reconstruction becomes practical.NeuroImage, 54(2), 2011

    Marco Reisert, Irina Mader, Constantin Anastasopoulos, Matthias Weigel, Susanne Schnell, and Valerij Kiselev. Global fiber reconstruction becomes practical.NeuroImage, 54(2), 2011. 2

    2011

  29. [29]

    Current challenges and opportunities for tractography

    Francois Rheault, Philippe Poulin, Alex Valcourt Caron, Etienne St-Onge, Kurt G Schilling, Laurent Petit, Flavio Dell’Acqua, Alexander Leemans, and Maxime Descoteaux. Current challenges and opportunities for tractography. Handbook of Diffusion MR Tractography, 2025. 2, 3

    2025

  30. [30]

    Prevalence of white matter pathways com- ing into a single white matter voxel orientation: The bottle- neck issue in tractography.Human Brain Mapping, 43(4),

    Kurt G Schilling, Chantal MW Tax, Francois Rheault, Ben- nett A Landman, Adam W Anderson, Maxime Descoteaux, and Laurent Petit. Prevalence of white matter pathways com- ing into a single white matter voxel orientation: The bottle- neck issue in tractography.Human Brain Mapping, 43(4),

  31. [31]

    Sift: Spherical-deconvolution in- formed filtering of tractograms.NeuroImage, 67, 2013

    Robert E Smith, Jacques-Donald Tournier, Fernando Cala- mante, and Alan Connelly. Sift: Spherical-deconvolution in- formed filtering of tractograms.NeuroImage, 67, 2013. 6

    2013

  32. [32]

    Bundleseg: A versatile, reliable and reproducible approach to white matter bundle segmentation

    Etienne St-Onge, Kurt G Schilling, and Francois Rheault. Bundleseg: A versatile, reliable and reproducible approach to white matter bundle segmentation. InInternational Work- shop on Computational Diffusion MRI. Springer, 2023. 5

    2023

  33. [33]

    Track-to-learn: A general framework for tractography with deep reinforcement learn- ing.Medical Image Analysis, 72, 2021

    Antoine Th ´eberge, Christian Desrosiers, Maxime De- scoteaux, and Pierre-Marc Jodoin. Track-to-learn: A general framework for tractography with deep reinforcement learn- ing.Medical Image Analysis, 72, 2021. 2, 5

    2021

  34. [34]

    TractOracle: towards an anatomically-informed re- ward function for RL-based tractography

    Antoine Th ´eberge, Maxime Descoteaux, and Pierre-Marc Jodoin. TractOracle: towards an anatomically-informed re- ward function for RL-based tractography. InProceedings of the International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, 2024. 1, 2, 6

    2024

  35. [35]

    Improved probabilistic streamlines tractography by 2nd order integration over fibre orientation distributions

    J Donald Tournier, Fernando Calamante, Alan Connelly, et al. Improved probabilistic streamlines tractography by 2nd order integration over fibre orientation distributions. In Proceedings of the International Society for Magnetic Reso- nance in Medicine. Stockholm, 2010. 1, 2, 6

    2010

  36. [36]

    Mrtrix3: A fast, flexible and open software framework for medical image processing and visualisation.NeuroImage, 202, 2019

    J Donald Tournier, Robert Smith, David Raffelt, Rami Tab- bara, Thijs Dhollander, Maximilian Pietsch, Daan Christi- aens, Ben Jeurissen, Chun-Hung Yeh, and Alan Connelly. Mrtrix3: A fast, flexible and open software framework for medical image processing and visualisation.NeuroImage, 202, 2019. 6

    2019

  37. [37]

    The WU-Minn Human Con- nectome Project: an Overview.NeuroImage, 80, 2013

    David C Van Essen, Stephen M Smith, Deanna M Barch, Timothy EJ Behrens, Essa Yacoub, Kamil Ugurbil, Wu- Minn HCP Consortium, et al. The WU-Minn Human Con- nectome Project: an Overview.NeuroImage, 80, 2013. 5

    2013

  38. [38]

    Attention is all you need.Advances in Neural Information Processing Systems, 30, 2017

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need.Advances in Neural Information Processing Systems, 30, 2017. 4

    2017

  39. [39]

    Noise reduction in magni- tude diffusion-weighted images using spatial similarity and diffusion redundancy.Magnetic Resonance Imaging, 118,

    Liming Yang and Yuanjun Wang. Noise reduction in magni- tude diffusion-weighted images using spatial similarity and diffusion redundancy.Magnetic Resonance Imaging, 118,

  40. [40]

    Think deep in the trac- tography game: deep learning for tractography computing and analysis.Brain Structure and Function, 230(6), 2025

    Fan Zhang, Antoine Th ´eberge, Pierre-Marc Jodoin, Maxime Descoteaux, and Lauren J O’Donnell. Think deep in the trac- tography game: deep learning for tractography computing and analysis.Brain Structure and Function, 230(6), 2025. 2 10

    2025