[Paper Review] Mitigating Gyral Bias in Cortical Tractography via Asymmetric Fiber Orientation Distributions

Ye Wu, Yoonmi Hong+3 more|arXiv (Cornell University)|Apr 6, 2019

Advanced Neuroimaging Techniques and ApplicationsMedicine36 references3 citations

TL;DR

This paper proposes a global, convex optimization framework for estimating asymmetric fiber orientation distribution functions (AFODFs) to reduce gyral bias in cortical tractography. By enforcing fiber continuity across voxels and capturing subvoxel asymmetry, AFODFs enable streamlines to bend more naturally into gyral sulci, significantly improving cortico-cortical connectivity and consistency across 3T and 7T MRI scans.

ABSTRACT

Diffusion tractography in brain connectomics often involves tracing axonal trajectories across gray-white matter boundaries in gyral blades of complex cortical convolutions. To date, gyral bias is observed in most tractography algorithms with streamlines predominantly terminating at gyral crowns instead of sulcal banks. This work demonstrates that asymmetric fiber orientation distribution functions (AFODFs), computed via a multi-tissue global estimation framework, can mitigate the effects of gyral bias, enabling fiber streamlines at gyral blades to make sharper turns into the cortical gray matter. We use ex-vivo data of an adult rhesus macaque and in-vivo data from the Human Connectome Project (HCP) to show that the fiber streamlines given by AFODFs bend more naturally into the cortex than the conventional symmetric FODFs in typical gyral blades. We demonstrate that AFODF tractography improves cortico-cortical connectivity and provides highly consistent outcomes between two different field strengths (3T and 7T).

Motivation & Objective

• Address the persistent issue of gyral bias in diffusion tractography, where streamlines preferentially terminate at gyral crowns instead of sulcal banks.
• Overcome the limitations of symmetric fiber orientation distribution functions (FODFs), which fail to represent complex, curved axonal trajectories in superficial white matter.
• Develop a method that improves cortical tractography without relying on high-resolution imaging or T1-weighted anatomical priors.
• Enable more accurate and consistent mapping of long-range cortico-cortical connections across different MRI field strengths.

Proposed method

• Formulate a multi-tissue global estimation framework that models fiber orientation distributions as asymmetric, capturing subvoxel fiber geometry in gyral blades.
• Enforce fiber continuity across neighboring voxels by minimizing the difference between incoming and outgoing fiber orientations (e.g., u and -u), ensuring physically plausible streamline trajectories.
• Construct a convex optimization problem to estimate AFODFs simultaneously across all voxels, avoiding initialization from symmetric FODFs and ensuring global convergence.
• Use diffusion-weighted imaging data directly to estimate AFODFs, without requiring precomputed symmetric FODFs or external anatomical constraints.
• Incorporate spatial regularization via inter-voxel continuity to resolve ambiguous fiber configurations in regions with fanning and bending trajectories.
• Apply spherical deconvolution principles but extend them to asymmetric FODFs by modeling distinct fiber polarity responses in local neighborhoods.

Experimental results

Research questions

• RQ1 Can asymmetric fiber orientation distribution functions (AFODFs) reduce gyral bias in cortical tractography compared to symmetric FODFs?
• RQ2 Does the proposed global, convex optimization framework for AFODF estimation improve the accuracy of fiber streamline trajectories in gyral blades?
• RQ3 To what extent do AFODF-based tractography results improve cortico-cortical connectivity and consistency across different MRI field strengths (3T vs. 7T)?
• RQ4 Can AFODFs better represent subvoxel fiber configurations in superficial white matter where complex fiber geometries (e.g., fanning, bending) occur?
• RQ5 Does the method maintain robustness and reproducibility without relying on high-resolution data or T1-weighted anatomical priors?

Key findings

• AFODF-based tractography significantly reduces gyral bias, with streamlines making sharper turns into sulcal banks rather than terminating prematurely at gyral crowns.
• Over 95% of streamlines initiated from cortical gray matter voxels successfully reach other cortical gray matter regions, indicating improved cortico-cortical connectivity.
• The method achieves high consistency in connectivity matrices between 3T and 7T MRI scans, with ICI values of 0.94 for AFODFs compared to lower values for symmetric FODFs.
• AFODF tractography produces fiber density maps that more closely align with histological expectations, showing balanced coverage at both gyral and sulcal WM-GM boundaries.
• The global, convex optimization framework enables robust, initialization-free AFODF estimation that outperforms non-convex, symmetric FODF-based methods in tractography accuracy.
• The method demonstrates reproducibility across different field strengths without requiring specialized high-resolution scanning protocols or anatomical priors.