基于Watersnake的视神经纤维的自动识别

Automatic Detection of Optic Nerve Fiber Based on Watersnake

在视神经横切面图像中,将每个神经纤维的边界精确地分割出来,是视神经形态分析的重要环节。由于神经纤维数量巨大、形态各异,单纯采用一种分割方法无法获得满意效果。采用基于Watersnake的分割方法,整合Watershed和Snake在图像分割方面的优势,分两步进行纤维分割。首先由Watershed获取纤维的粗轮廓,在此基础上由Snake经曲线演化获得纤维的精细轮廓。实验结果表明,Watersnake能较好地识别神经纤维的内外轮廓,为后续的神经纤维形态分析奠定基础。

The segmentation of nerve fiber in the transverse section of optic nerve is an important foundation of the optic nerve morphometry. A single segmentation method is not efficient because of the large amount and various shape of nerve fiber. The approach of watersnake was introduced in this investigation which integrated watershed and snake. Firstly, the coarse boundary of fiber was extracted by watershed, and then the fine boundary was obtained by snake. The experiment results illustrate that watersnake can efficiently identify the inner and outer boundaries of nerve fibers, which is the basis of nerve fiber morphometry.