Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

Implicit Differentiation01:25

Implicit Differentiation

In classical mechanics, motion is often described through relationships between spatial coordinates and time. A car moving along a straight highway with constant acceleration serves as a simple case where velocity is an explicit function of time. This scenario results in a linear equation, enabling straightforward analysis using basic differentiation techniques.In contrast, a satellite in circular orbit follows a path defined by an implicit function. The position of the satellite is constrained...
Logarithmic Differentiation01:28

Logarithmic Differentiation

When a car’s weight and driving forces act on a tire, they impose an external load on the rubber material. This load is resisted internally by forces distributed throughout the tire structure, which are defined as stress. The resulting deformation of the rubber due to this stress is quantified as strain. The relationship between stress and strain governs how the tire deforms under load and is central to understanding its mechanical response during operation.Rubber exhibits a nonlinear...
Implicit Differentiation with Partial Derivatives01:27

Implicit Differentiation with Partial Derivatives

Implicit differentiation with partial derivatives is used when a relationship between two variables is defined implicitly rather than explicitly. Instead of solving one variable in terms of the other, the variables remain connected through a single equation. In this setting, one variable is treated as depending on the other, and differentiation is applied directly to the entire relation.To differentiate an implicit relation, the chain rule is applied to every term in the equation. Because one...
Maximizing the Directional Derivative01:25

Maximizing the Directional Derivative

The directional derivative is a central concept in multivariable calculus that describes how a function changes at a given point when moving in a specified direction. This direction is represented by a unit vector, ensuring that only the orientation influences the rate of change. By varying the direction, different rates of change can be observed, demonstrating that the directional derivative depends strongly on the chosen direction.The directional derivative is computed using the gradient...
Forced Transdifferentiation01:28

Forced Transdifferentiation

Transdifferentiation, also known as lineage reprogramming, was first discovered by Selman and Kafatos in 1974 in silkmoths. They observed that the moths’ cuticle-producing cells transformed into salt-producing cells. Many such cases of natural transdifferentiation occur in organisms. In humans, pancreatic alpha cells can become beta cells. In newts, the loss of the eye’s lens causes the pigmented epithelial cells to transdifferentiate into the lens cells.
Artificial transdifferentiation occurs...
Implicit Differentiation: Problem Solving01:29

Implicit Differentiation: Problem Solving

Curves defined implicitly, where variables cannot be separated algebraically, require specialized techniques for analysis. The conchoid of Nicomedes exemplifies such a case. Its equation links x and y in a way that prevents isolation of one variable, making implicit differentiation essential to determine the slope and behavior at any point on the curve.The implicit form of the conchoid can be expressed as:To differentiate this equation, y is treated as a function of x, and the chain rule is...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

New Growth, New Opportunities.

Journal of medical imaging (Bellingham, Wash.)·2025
Same author

White matter hyperintensities and relapse risk in late-life depression.

Journal of affective disorders·2025
Same author

Unsupervised discovery of clinical disease signatures using probabilistic independence.

Journal of biomedical informatics·2025
Same author

Multi-contrast computed tomography atlas of healthy pancreas with dense displacement sampling registration.

Journal of medical imaging (Bellingham, Wash.)·2025
Same author

The effect of Alzheimer's disease genetic factors on limbic white matter microstructure.

Alzheimer's & dementia : the journal of the Alzheimer's Association·2025
Same author

Segment Anything Model (SAM) for Digital Pathology: Assess Zero-shot Segmentation on Whole Slide Imaging.

IS&T International Symposium on Electronic Imaging·2025
Same journal

Deep learning‑based cell type prediction in lung tissue from brightfield histology using CODEX-derived labels.

Proceedings of SPIE--the International Society for Optical Engineering·2026
Same journal

Reconstructing physiological signals from fMRI across the adult lifespan.

Proceedings of SPIE--the International Society for Optical Engineering·2026
Same journal

Axially Swept Light-Sheet Microscopy using scattering and fluorescence contrast mechanisms.

Proceedings of SPIE--the International Society for Optical Engineering·2026
Same journal

Analytic Bounds on GAMLSS Model Variability of Normative White Matter Brain Charts.

Proceedings of SPIE--the International Society for Optical Engineering·2026
Same journal

Personalized White Matter Bundle Segmentation for Early Childhood.

Proceedings of SPIE--the International Society for Optical Engineering·2026
Same journal

Impact of reconstruction kernel variability on segmentation consistency in low-dose thoracic CT.

Proceedings of SPIE--the International Society for Optical Engineering·2026
See all related articles
  1. Home
  3. Fully Differentiable Dmri Streamline Propagation In Pytorch.
  1. Home
  3. Fully Differentiable Dmri Streamline Propagation In Pytorch.

Related Experiment Videos

Fully Differentiable dMRI Streamline Propagation in PyTorch.

Jongyeon Yoon1, Elyssa M McMaster2, Michael E Kim1

  • 1Department of Computer Science, Vanderbilt University, Nashville, TN, USA.

Proceedings of Spie--The International Society for Optical Engineering
|June 10, 2026

View abstract on PubMed

Summary
This summary is machine-generated.

This article introduces a new way to map brain white matter pathways using a fully differentiable computer program. By making the process of tracing these pathways compatible with modern artificial intelligence tools, researchers can now build more powerful models for studying brain structure. This approach maintains high accuracy while allowing for seamless integration into advanced machine learning workflows.

Keywords:
Diffusion MRIPyTorchStreamline PropagatorTractographydeep learningwhite matter pathwaysneural networksneuroimaging software

Frequently Asked Questions

Related Experiment Videos

Area of Science:

  • Neuroimaging research within diffusion MRI connectivity analysis
  • Computational neuroscience and differentiable programming methods

Background:

No prior work had resolved the limitation of non-differentiable streamline propagation in existing brain imaging pipelines. Diffusion MRI provides a unique window into the microstructural organization of biological tissues. Researchers utilize this imaging modality to map connectivity and estimate various macrostructural features. Tractography emerged as a powerful tool for visualizing white matter pathways within the human brain. Most current methods rely on procedural propagators or global energy minimization techniques. These traditional approaches often lack the necessary mathematical properties for modern deep learning integration. That uncertainty drove the need for a framework that supports end-to-end gradient flow. This paper addresses the gap by introducing a fully differentiable solution for streamline generation.

Purpose Of The Study:

The aim of this study is to develop a fully differentiable solution for streamline propagation in diffusion MRI. Researchers sought to overcome the limitations of existing methods that prevent integration into modern deep learning frameworks. The team identified that current tractography approaches are often non-differentiable, which restricts their utility in end-to-end learning tasks. This gap motivated the creation of a framework that retains numerical fidelity with established algorithms. The authors intended to provide a tool that supports gradient flow throughout the entire propagation process. They aimed to enable deeper integration of tractography into complex neural network workflows. By addressing this technical hurdle, the study seeks to facilitate a new category of macrostructural reasoning. The researchers focused on building a robust system that maintains scientific rigor while enhancing computational flexibility.

The researchers propose a PyTorch-engineered propagator that eliminates components blocking gradient flow. Unlike traditional procedural methods that are non-differentiable, this new solution allows for end-to-end learning by maintaining numerical fidelity with established streamline algorithms.

The authors utilize PyTorch, a popular deep learning library, to implement their propagator. This tool is essential for creating a framework where gradient information can flow through the entire tractography process, unlike standard software packages that lack this capability.

A fully differentiable approach is necessary because existing tractography methods are non-differentiable, which prevents their integration into end-to-end learning pipelines. This technical requirement allows researchers to optimize tractography parameters directly within neural network training workflows.

The framework relies on PyTorch-based tensors to represent streamline coordinates and directions. This data structure plays a central role by ensuring that every step of the propagation process remains mathematically connected to the input diffusion data.

The researchers measure the numerical fidelity of their method by comparing it against standard, non-differentiable streamline propagators. They report that their solution matches these established benchmarks while providing the added benefit of full differentiability.

The authors claim that this framework lays the foundation for a new category of macrostructural reasoning. They propose that this advancement will lead to more computationally robust and scientifically rigorous analysis of white matter pathways.

Main Methods:

The team designed a streamline propagator entirely within the PyTorch environment. This approach focuses on removing any operations that interrupt the flow of gradients during computation. The authors translated standard procedural logic into a series of differentiable tensor operations. They verified the accuracy of their implementation by benchmarking against a leading conventional algorithm. The design ensures that the propagation process remains mathematically consistent with established neuroimaging standards. Researchers utilized synthetic and real-world diffusion data to validate the robustness of the framework. This methodology avoids the use of non-differentiable black-box components found in older software. The final architecture supports direct backpropagation through the entire streamline generation pipeline.

Main Results:

The primary finding shows that the new method matches the performance of standard, non-differentiable propagators. This result confirms that numerical fidelity is maintained despite the transition to a differentiable framework. The authors report that their PyTorch-engineered solution successfully eliminates all components that previously blocked gradient flow. By enabling end-to-end learning, the framework allows for the optimization of tractography parameters within deep learning models. The study provides evidence that this approach is both computationally robust and scientifically rigorous. Comparisons indicate that the output streamlines are consistent with those generated by traditional procedural techniques. The researchers demonstrate that their tool integrates seamlessly into existing deep learning workflows. These results establish a new standard for differentiable macrostructural reasoning in brain connectivity analysis.

Conclusions:

The authors demonstrate that their new framework maintains numerical fidelity compared to established streamline algorithms. This solution allows for the seamless incorporation of tractography into complex deep learning architectures. By removing barriers to gradient flow, the researchers enable new forms of macrostructural reasoning. The study confirms that differentiable propagation matches the performance of standard procedural methods. This development provides a foundation for more robust computational analysis of brain connectivity. The team suggests that their approach enhances the scientific rigor of tractography workflows. Future applications may benefit from the deeper integration of these pathways into neural network training. These findings highlight the potential for advancing neuroimaging through differentiable programming techniques.