Segmentation of the Corpus Callosum from Diffusion Tensor Images Based on Improved Marker Watershed Algorithm

李婷,吕晓琪,谷宇,付龙,郝娟
DOI: https://doi.org/10.13437/j.cnki.jcr.2017.03.043
2017-01-01
Abstract:Objective To segment out the corpus callosum from diffusion tensor images more quickly and accurately,a new improved method based on watershed algorithm is proposed.Methods The diffusion tensor imaging data of corpus callosum is processed to get the fractional anisotropy image.Then the Otsu is used as the fitness function of the particle swarm optimization to get the best threshold which is used for watershed segmentation algorithm.Results The experimental results show that the new algorithm is effective to filter out the spurious local minima of Traditional watershed algorithm caused by noise.It can restrain the over-segmentation and obtain good effect of the segmentation of the corpus callosum.Conclusion Compared with the traditional marker watershed algorithm,the new algorithm produces significant increases in speed and accuracy of segmentation.
What problem does this paper attempt to address?