Geodesic fiber tracking in white matter using activation function

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Medical Physics

Background:

• Geodesic ray-tracing is effective for white matter tract reconstruction.
• Diffusion tensor imaging (DTI) metrics significantly impact reconstruction outcomes.
• Existing methods face challenges with noise, fiber curvature, and crossings.

Purpose of the Study:

• To propose improved metrics and algorithmic modifications for white matter tract reconstruction.
• To enhance the robustness and accuracy of geodesic ray-tracing.
• To better handle fiber crossings and noise in diffusion tensor imaging.

Main Methods:

• Rescaling diffusion tensor metrics to increase geodetic cost in isotropic regions.
• Applying conformal changes to preserve fiber angles during tracking.
• Enhancing noise robustness and utilizing fourth-order tensor data for complex crossings.

Main Results:

• A method for selecting conformal metrics scaled by tensor anisotropy using logistic functions.
• A hybrid ray-tracing approach to prevent geodesic deviation.
• Utilizing diagonal projections of 4th order tensors for fiber tracking in crossing regions.

Conclusions:

• Successfully implemented and tested new algorithms on synthetic and real data.
• Demonstrated improved performance compared to previous approaches.
• Validated the effectiveness of proposed metrics and methods for white matter tractography.