Automatic fiber bundle segmentation in massive tractography datasets using a multi-subject bundle atlas.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Medical Image Analysis

Background:

• Diffusion MRI (dMRI) tractography generates complex datasets of white matter fiber bundles.
• Accurate segmentation of these bundles is crucial for understanding brain connectivity and neurological disorders.
• Existing methods often struggle with inter-subject variability in white matter pathways.

Purpose of the Study:

• To develop and validate an automated method for segmenting white matter fiber bundles from large dMRI tractography datasets.
• To create a robust multi-subject bundle atlas representing population-level white matter organization.
• To improve the accuracy and efficiency of white matter tract segmentation.

Main Methods:

• A two-level clustering strategy (intra-subject and inter-subject) was employed to derive a multi-subject bundle atlas.
• The atlas models brain white matter organization using generic fiber bundles representative of the population.
• Atlas bundles were constructed from manually labeled inter-subject clusters to capture pathway variability.

Main Results:

• The developed atlas comprises 36 deep and 47 superficial white matter bundles per hemisphere.
• The method successfully segmented deep white matter bundles in an independent database of 20 brains.
• Most superficial white matter bundles were also accurately segmented in 10 subjects.

Conclusions:

• The proposed atlas-based method provides an effective approach for automatic segmentation of white matter fiber bundles.
• This technique demonstrates high accuracy in segmenting both deep and superficial white matter pathways.
• The method holds promise for advancing research in neuroanatomy and clinical neuroscience.