Delineating white matter structure in diffusion tensor MRI with anisotropy creases.

Area of Science:

• Neuroscience
• Medical Imaging
• Computer Vision

Background:

• Geometric models of white matter architecture are crucial for diffusion tensor imaging (DTI) applications.
• Fiber tractography is the predominant method for constructing these models.
• Derivatives of diffusion anisotropy offer a potential alternative for specific neuroscientific analyses.

Purpose of the Study:

• To introduce and define anisotropy creases as features of locally extremal tensor anisotropy in tensor fields.
• To propose anisotropy creases as a basis for extracting a skeleton of major white matter pathways.
• To develop and present an algorithm for extracting anisotropy crease surfaces.

Main Methods:

• Extended the concept of ridges and valleys from computer vision to tensor fields.
• Defined anisotropy creases mathematically as loci where the anisotropy gradient is orthogonal to Hessian eigenvectors.
• Developed a crease extraction algorithm generating polygonal models, simplified using connected-component analysis.

Main Results:

• Demonstrated the extraction of anisotropy creases from measured diffusion MRI data.
• Visualized anisotropy creases in conjunction with fiber tractography.
• Confirmed the anatomic relevance of anisotropy creases, with ridges aligning with fiber tract interiors and valleys with tract interfaces.

Conclusions:

• Anisotropy creases provide a compelling alternative to fiber tractography for certain white matter analysis tasks.
• The proposed method generates high-quality models of white matter pathways.
• Anisotropy creases offer valuable insights into white matter architecture and connectivity.