Probabilistic Mapping and Automated Segmentation of Human Brainstem White Matter Bundles

Area of Science:

• Neuroimaging and Connectomics
• Diffusion MRI Tractography
• Computational Neuroscience

Background:

• Brainstem white matter bundles are critical for vital functions but challenging to map due to their small size and complex morphology.
• Existing neuroimaging methods lack automated tracing tools for these essential brainstem structures.
• Accurate mapping is vital for understanding brainstem connectomics and neurological disorders.

Purpose of the Study:

• To develop an automated method for segmenting key white matter bundles in the rostral brainstem.
• To validate the tool's robustness across different imaging protocols and sample types.
• To investigate the utility of automated brainstem mapping in neurological diseases and patient prognostics.

Main Methods:

• Leveraged diffusion MRI tractography to create the BrainStem Bundle Tool (BSBT).
• Employed a convolutional neural network on a custom probabilistic fiber map for automated segmentation of eight brainstem white matter bundles.
• Validated BSBT using in vivo and ex vivo human brain scans with histological correlation.

Main Results:

• BSBT successfully performed automated segmentation of eight key rostral brainstem white matter bundles.
• Demonstrated robustness across various diffusion MRI acquisition protocols.
• Identified distinct alterations in brainstem white matter bundles in Alzheimer's disease, Parkinson's disease, and traumatic brain injury cohorts.
• Showcased prognostic utility in a longitudinal study of coma recovery.

Conclusions:

• The BrainStem Bundle Tool (BSBT) offers a robust, automated solution for mapping brainstem white matter.
• BSBT facilitates large-scale analysis of brainstem white matter in neurological disorders.
• This tool holds significant potential for advancing our understanding of brainstem function and its role in disease and recovery.