Automated Segmentation of Brainstem and Subcortical White Matter: Mapping the Deep Tegmental Core with BundleParc

Area of Science:

• Neuroimaging
• Neuroanatomy
• Computational Neuroscience

Background:

• Diffusion MRI is vital for mapping white matter, but automated methods often overlook brainstem and subcortical pathways.
• These deep pathways are critical for basal ganglia, cerebellar, limbic, sensory, and homeostatic functions, yet are underrepresented in connectomics.
• Existing methods lack comprehensive segmentation for these intricate white matter tracts.

Purpose of the Study:

• To develop an automated pipeline for segmenting and parceling 97 subcortical and brainstem white matter pathways.
• To adapt the BundleParc architecture for direct segmentation and along-tract parcellation of deep brain structures.
• To create a valuable resource for studying these pathways across development, aging, and disease.

Main Methods:

• Adapted the BundleParc architecture into an automated pipeline using Human Connectome Project diffusion MRI data.
• Employed anatomy-guided tractography, strict inclusion/exclusion criteria, outlier filtering, and manual quality assurance.
• Operated directly on native-space fiber orientation distributions for accurate tract recovery.

Main Results:

• Successfully segmented and parcelled 97 subcortical and brainstem white matter pathways with high anatomical accuracy.
• Demonstrated robust generalization across diverse datasets including development, aging, and neurodegenerative disease cohorts.
• Maintained consistent performance despite variations in spatial resolution and angular sampling.

Conclusions:

• The developed automated pipeline effectively maps complex subcortical and brainstem white matter pathways.
• This resource enables advanced research into the role of these pathways in various neurological conditions and development.
• The released tools and data facilitate future studies in neuromodulation and brain connectomics.