Automatic Vessel Extraction Using Particle Swarming Optimization for 3D Medical Images

Abstract

Background: Blood vessel segmentation plays an important role in medical image analysis. Modern blood vessel segmentation algorithms attempts to attempts to increase patient safety by providing better diagnosis and support to more accurate medical decisions. Methods: In 3D image processing techniques leads to an emerging area, and voxel classification. Most of the voxel classification algorithms are a manual classification. This work introduces a novel fully automatic blood vessel segmentation algorithm from 3D images using Hessian-based multi-scale filters (Frangi's filter) and Chan-Vese model with level-set framework. Parameters of Frangi's filter are adjusted by means of an evolutionary computation method, particle swarm optimization (PSO). 3D synthetic and real CTA clinical image database is used to test the proposed algorithm and show a correct voxel classification. Conclusion: The proposed algorithm shows results that are more accurate. Keywords: Particle swarm optimization, Chan-Vese model, level-set, Hessian-based multi-scale filter, 3D medical images, blood vessel segmentation.