Automated facial-vestibulocochlear nerve complex identification based on data-driven tractography clustering.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Medical Image Analysis

Background:

• Tractography of the facial-vestibulocochlear nerve complex (FVN) is challenging due to its small size and complex anatomy.
• Manual identification requires significant expertise, time, and resources, limiting its accessibility.

Purpose of the Study:

• To develop and validate the first automated pipeline for identifying the FVN.
• To overcome the limitations of manual FVN tractography and identification.

Main Methods:

• Creation of an FVN template using multishell diffusion MRI data and spectral fiber clustering.
• Development of a data-driven pipeline involving fiber clustering, atlas creation, and spectral embedding for automatic FVN identification.
• Validation on healthy and tumor patient datasets from different acquisition sites.

Main Results:

• The automated method achieved ideal spatial overlap and visualization compared to expert manual identification.
• Successful application to tumor patient data, with identified FVNs aligning with intraoperative findings.
• Demonstrated robustness across different datasets and acquisition parameters.

Conclusions:

• The proposed automated pipeline effectively identifies the FVN, offering a reliable and efficient alternative to manual methods.
• This approach has significant potential for clinical applications, particularly in neurosurgical planning and research involving the FVN.
• The study establishes a novel methodology for automated neuroanatomical structure identification using diffusion MRI tractography.