White matter fiber tract segmentation in DT-MRI using geometric flows

Area of Science:

• Medical Imaging
• Computational Anatomy
• Neuroscience

Background:

• Diffusion Tensor Magnetic Resonance Imaging (DT-MRI) is crucial for visualizing white matter tracts.
• Accurate segmentation of fiber tracts is essential for quantitative analysis and clinical applications.
• Existing methods may require tensor field regularization, adding complexity.

Purpose of the Study:

• To present a novel 3D geometric flow for segmenting the main core of fiber tracts in DT-MRI.
• To develop a method that does not require a regularized tensor field.
• To enable quantitative diffusion measures, white matter registration, and surgical planning.

Main Methods:

• A 3D geometric flow utilizing front propagation to fill entire fiber tracts.
• Propagation speed is determined by diffusion tensor similarity between surface voxels and neighbors.
• Level set implementation ensures stable surface evolution and handles topological changes.
• Intrinsic surface smoothing based on minimal principal curvature eliminates the need for tensor field regularization.

Main Results:

• Successfully segmented the main core of fiber tracts using the proposed geometric flow.
• The method demonstrated stable and accurate surface evolution with automatic smoothing.
• Achieved segmentation without requiring a pre-regularized tensor field.

Conclusions:

• The presented 3D geometric flow offers an effective approach for fiber tract segmentation in DT-MRI.
• This method facilitates quantitative diffusion analysis, white matter registration, and surgical planning.
• The automatic smoothing mechanism simplifies the segmentation process.