Subject-Specific Automatic Reconstruction of White Matter Tracts

Area of Science:

• Neuroimaging
• Neurosurgery
• Medical image analysis

Background:

• MRI-based tractography is currently underutilized in brain tumor surgery due to inconsistent methods and lack of standardized systems.
• Standardization is crucial for objective validation and clinical integration of tractography techniques.

Purpose of the Study:

• To evaluate the robustness and reliability of a novel tractography pipeline in healthy volunteers.
• To establish a standardized, automated approach for generating subject-specific white matter tract information.

Main Methods:

• Utilized a deep learning convolutional network for automated segmentation of regions of interest.
• Employed an evidence-based tractography protocol with post-processing filtering on data from 136 healthy volunteers (Human Connectome Project).
• Assessed robustness using cross-validation and reliability through repeated scans and test-retest analysis.

Main Results:

• The tractography pipeline demonstrated high robustness with consistent probability maps and no significant outliers.
• Individual-level reliability was high, approaching the upper limit in test-retest assessments.
• The method successfully generated robust and reliable subject-specific white matter tracts.

Conclusions:

• The developed tractography pipeline is robust and reliable for mapping white matter tracts in healthy subjects.
• This standardized approach is a critical step towards validating and implementing tractography in clinical neurosurgery, particularly for brain tumor surgery.
• The findings support the potential for improved surgical planning and patient outcomes through advanced neuroimaging techniques.