Data-Driven Decomposition of Precisely Aligned Fiber-Orientation-Preserving Probabilistic Tractography Connectomes with Non-Negative Matrix Factorization

The major goals of the Human Connectome Project (HCP) were to characterize the structural and functional connectivity of the human brain. Although there has been tremendous work done for functional connectivity, including application of a variety of data driven decomposition approaches, structural connectivity from the HCP has seen less attention. This work presents a collection of methodological advances that, in combination, enable similar approaches to be applied to diffusion tractography data. Here, we use multi-modal non-linear volumetric alignment with FSL's MultiMOdal Registration Framework (MMORF) to precisely align white matter tracts and globular gray matter nuclei, use the White matter Hierarchical Modelling (WHIM) method to align voxel-wise fiber orientations, enable efficient GPU-accelerated computation and storage of full resolution probabilistic gray matter x white matter x fiber sparse tractography matrices in precise alignment across individuals using the aforementioned MMORF and WHIM alignments together with areal-feature-based cortical surface registration. These tractography dense connectomes can be averaged across individuals and sparse, computationally efficient non-negative matrix factorization (NMF) can be applied to perform data-driven decompositions into white matter tracts and gray matter projection territories. The primary outputs of this effort to date are 1) A set of HCP software pipelines enabling such analyses, 2) A new unbiased HCP-derived volumetric standard space that is representative of the young adult brain and follows ACPC orientation conventions across individuals. 3) Reproducible NMF decompositions of HCP young adult tractography data into white matter tracts and gray matter projection territories, which could be projected using non-negative dual regression back to individuals. These data represent a starting point for a new generation of data-driven atlasing efforts for human white matter, and these methods can be applied to study the white matter in development, aging, and brain diseases.

Authors

Institutions

Publication Details

Journal
bioRxiv (Cold Spring Harbor Laboratory)
Published
2026-11-19
DOI
https://doi.org/10.64898/2026.11.08.765179
Primary Topic
Advanced Neuroimaging Techniques and Applications
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Data-Driven Decomposition of Precisely Aligned Fiber-Orientation-Preserving Probabilistic Tractography Connectomes with Non-Negative Matrix Factorization

Matthew F. Glasser, Frederik Lange, Burke Q. Rosen, Saâd Jbabdi et al.
bioRxiv (Cold Spring Harbor Laboratory)
Advanced Neuroimaging Techniques and Applications
preprint

Data-Driven Decomposition of Precisely Aligned Fiber-Orientation-Preserving Probabilistic Tractography Connectomes with Non-Negative Matrix Factorization

Matthew F. Glasser, Frederik Lange, Burke Q. Rosen, Saâd Jbabdi, David C. Van Essen, Stamatios N. Sotiropoulos, Hossein Rafipoor, Timothy S. Coalson, Takuya Hayashi, Naofumi Yoshida, Alexander Zhou, Andrea Yang
preprint en

Abstract

The major goals of the Human Connectome Project (HCP) were to characterize the structural and functional connectivity of the human brain. Although there has been tremendous work done for functional connectivity, including application of a variety of data driven decomposition approaches, structural connectivity from the HCP has seen less attention. This work presents a collection of methodological advances that, in combination, enable similar approaches to be applied to diffusion tractography data. Here, we use multi-modal non-linear volumetric alignment with FSL's MultiMOdal Registration Framework (MMORF) to precisely align white matter tracts and globular gray matter nuclei, use the White matter Hierarchical Modelling (WHIM) method to align voxel-wise fiber orientations, enable efficient GPU-accelerated computation and storage of full resolution probabilistic gray matter x white matter x fiber sparse tractography matrices in precise alignment across individuals using the aforementioned MMORF and WHIM alignments together with areal-feature-based cortical surface registration. These tractography dense connectomes can be averaged across individuals and sparse, computationally efficient non-negative matrix factorization (NMF) can be applied to perform data-driven decompositions into white matter tracts and gray matter projection territories. The primary outputs of this effort to date are 1) A set of HCP software pipelines enabling such analyses, 2) A new unbiased HCP-derived volumetric standard space that is representative of the young adult brain and follows ACPC orientation conventions across individuals. 3) Reproducible NMF decompositions of HCP young adult tractography data into white matter tracts and gray matter projection territories, which could be projected using non-negative dual regression back to individuals. These data represent a starting point for a new generation of data-driven atlasing efforts for human white matter, and these methods can be applied to study the white matter in development, aging, and brain diseases.

bioRxiv (Cold Spring Harbor Laboratory)
University of Nottingham (GB), Washington University in St. Louis (US), University of Oxford (GB), RIKEN (JP), RIKEN Center for Biosystems Dynamics Research (JP)
Advanced Neuroimaging Techniques and Applications
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon