Three-dimensional skeletonization and symbolic description in vascular imaging: preliminary results

Area of Science:

• Medical Imaging
• Computational Anatomy
• Neurosurgery

Background:

• Complex anatomical structures, like the cerebral vascular tree, require advanced analysis methods.
• Current methods may not adequately capture the intricate morphology of vascular networks.

Purpose of the Study:

• To develop a general method for analyzing and describing tree-like structures, focusing on the cerebral vascular tree.
• To enable clinical applications in neurosurgery, particularly for planning intraoperative gestures.

Main Methods:

• Developed a 3D skeletonization method adapted for tubular forms with symbolic description.
• Utilized an iterative Dijkstra minimum cost spanning tree for branch-by-branch skeleton extraction.
• Tested the method on simulated data and clinical magnetic resonance imaging (MRI) datasets.

Main Results:

• The method demonstrated accuracy with a Haussdorff's distance < 17 voxels.
• Robustness was verified through 18 volume rotations, achieving a Dice's similarity coefficient > 70%.
• The symbolic description successfully simplified vascular networks into skeleton representations.

Conclusions:

• The symbolic description method effectively analyzes and interprets vascular networks from angiographic images.
• Provides a simplified skeleton representation and tree-like view of vascular information.
• Facilitates improved understanding and planning for neurosurgical interventions.