Globally optimized fiber tracking and hierarchical clustering -- a unified framework

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Medical Image Analysis

Background:

• Noninvasive characterization of human brain structural connectivity is crucial for understanding neural networks.
• Diffusion tensor imaging (DTI)-based fiber tractography is a key technique for mapping brain pathways.

Purpose of the Study:

• To present a novel fiber tractography technique integrating global optimization and hierarchical clustering.
• To improve the accuracy and robustness of mapping complex white matter pathways in the human brain.

Main Methods:

• Development of a globally optimized fiber tracking algorithm.
• Application of k-means clustering with modified Hubert statistic for fiber pathway partitioning.
• Iterative sampling, perturbation, and clustering within fiber bundles for optimal solutions.

Main Results:

• The proposed technique effectively partitions anatomically coherent fiber bundles.
• Global optimality enhances resistance to image artifacts and handles complex fiber structures.
• Hierarchical clustering naturally reconstructs and partitions multiple fiber bundles between regions.

Conclusions:

• The novel technique offers significant advantages over traditional streamline tractography for DTI.
• This method holds promise for clinical studies, especially in understanding structure-function relationships in the human brain.