Anatomy-to-tract mapping infers white matter pathways without diffusion streamline propagation.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Medical Image Analysis

Background:

• Diffusion tractography, essential for white matter mapping, is limited by diffusion MRI's signal-to-noise ratio and resolution.
• Current methods rely on streamline propagation, which can be inaccurate for complex fiber architectures like crossing or kissing fibers.

Purpose of the Study:

• To introduce Anatomy-to-Tract Mapping (ATM), a novel framework for generating white matter streamlines.
• To bypass the limitations of diffusion MRI by directly utilizing high-quality anatomical MRI data.
• To synthesize subject-specific and anatomically plausible streamlines, even in regions with complex fiber configurations.

Main Methods:

• ATM generates bundle-specific streamlines directly from T1-weighted MRI, eliminating the need for orientation field estimation or streamline propagation.
• The framework learns from paired T1-weighted and tractogram data to synthesize streamlines.
• Evaluation was performed on the TractoInferno dataset, comparing ATM against diffusion-based methods like MRtrix and SCIL atlas warping for 30 white matter bundles.

Main Results:

• ATM successfully generates anatomically plausible, subject-specific streamlines directly from T1-weighted MRI.
• The anatomy-driven approach robustly reconstructs complex fiber configurations, including crossing, kissing, and bending fibers.
• ATM demonstrated strong performance across multiple metrics, including bundle similarity, volume coverage, angular correlation, and geometric fidelity, outperforming diffusion-based methods.

Conclusions:

• Anatomy-to-Tract Mapping (ATM) offers a robust and accurate alternative for white matter streamline generation.
• By leveraging anatomical MRI, ATM overcomes key limitations of diffusion MRI-based tractography.
• This approach holds significant potential for advancing white matter mapping and understanding brain connectivity.