Boundary estimation of fiber bundles derived from diffusion tensor images

Area of Science:

• Medical Imaging
• Neuroscience
• Computational Anatomy

Background:

• Diffusion Tensor Imaging (DTI) enables non-invasive visualization of white matter tracts.
• Accurate delineation of fiber bundle boundaries is essential for neurosurgical planning.
• Current methods may lack precision in defining the extent of white matter structures.

Purpose of the Study:

• To present a novel iterative approach for estimating the boundaries of white matter fiber bundles using DTI data.
• To improve the accuracy of fiber bundle boundary determination for enhanced neurosurgical interventions.

Main Methods:

• An iterative algorithm utilizing the centerline of tracked fiber bundles between regions of interest (ROIs).
• Radial ray casting from the centerline, with 2D contour calculation at sampled points.
• Boundary point identification based on anisotropy and angle parameters along each ray.

Main Results:

• The novel method was tested on artificially generated DTI datasets.
• Performance was evaluated by varying parameters such as ray density and sampling rates.
• Achieved Dice Similarity Coefficients (DSC) ranging from 74.7% to 91.5%.

Conclusions:

• The presented iterative DTI method effectively estimates white matter fiber bundle boundaries.
• This technique provides valuable information for identifying high-risk structures in neurosurgery.
• The approach contributes to safer and more precise neurosurgical interventions.