Volume segmentation of susceptibility weighted images of the brain using a level set approach

An improved version of the variational level set algorithm was used for volume segmentation of susceptibility weighted images of the brain. Binary maps based on an intensity histogram were used as edge indicators in the level set evolution equation. The parameters of the evolution equation were optimized with a synthetic SWI phantom using a simulated annealing algorithm. The convergence condition for brain tissue segmentation was presented. The algorithm was applied to the mIP of the 3D magnitude images to define the brain contour, which was then used as the initialization contour in the volume segmentation in all slices. Robust volume segmentation was obtained and the signal loss in the peripheral regions of the brain was effectively reduced in the mIP display of the 3D SWI data. The level set algorithm provides a feasible solution for robust and fully automated volume segmentation for the display of 3D SWI of the brain.