Directional correlation characterization and classification of white matter tracts

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biomedical Engineering

Background:

• Understanding white matter (WM) tract architecture is crucial for neuroscience.
• Existing methods may not fully capture the directional properties of WM tracts.
• Quantitative indices are needed to systematically analyze WM tract morphology.

Purpose of the Study:

• To develop and validate a novel method for characterizing white matter (WM) tract architecture.
• To introduce directional correlation (DC) as a quantitative index for directional similarity in WM tracts.
• To establish a systematic classification routine for WM tracts using DC.

Main Methods:

• Utilized directional correlation (DC), the inner product of major eigenvectors of adjacent pixels, as a quantitative index.
• Employed a region-growing algorithm with a directional correlation threshold (DCt) to propagate areas.
• Tested the classification routine on in vivo mouse brain data.

Main Results:

• Individual WM tracts exhibit a high degree of directional similarity.
• Increasing the DC threshold (DCt) effectively distinguished neighboring WM tracts.
• The proposed algorithm successfully recognized WM tracts in the mouse brain dataset.

Conclusions:

• Directional correlation (DC) provides a robust measure for assessing WM tract directional similarity.
• The developed classification routine, using DC and DCt, enables accurate WM tract recognition.
• This approach offers a systematic way to study WM tract architecture in neuroimaging data.