Evaluation of tractogram filtering methods using human-like connectome phantoms.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biomedical Engineering

Background:

• Diffusion MRI tractography reconstructs white matter pathways but often generates spurious connections.
• Microstructure-informed filtering algorithms exist to enhance streamline plausibility, but comparative performance is underexplored.

Purpose of the Study:

• To systematically evaluate streamline filtering and post-processing algorithms using simulated connectome phantoms.
• To assess the impact of white matter architecture complexity on filtering algorithm performance.

Main Methods:

• Developed a framework for generating realistic connectome phantoms with brain-like fiber architectures.
• Evaluated streamline filtering algorithms including SIFT, COMMIT, and LiFE on these phantoms.
• Quantified performance using F-measure for deterministic and probabilistic tractography.

Main Results:

• All tested filtering methods improved connectome accuracy.
• Filtering provided significant improvements for simple fiber bundles but modest gains for complex architectures.
• Probabilistic tractography and weighted connectomes benefited most from filtering.

Conclusions:

• Streamline filtering algorithms enhance tractography accuracy, with efficacy dependent on white matter complexity.
• Further advancements in tractography and filtering are needed for precise connectome mapping.
• Filtering shows particular promise for probabilistic tractography and weighted connectomes.