周围神经图像分割算法研究

A Study on Peripheral Nerve Image Segmentation Algorithm

提出一种交互式神经组织CT图像分割算法,本算法基于交互式聚类图像分割算法,针对CT图像中神经组织的特点加以改进,采用自适应空间邻域信息混合高斯模型ASIGMM进行建模,从而能够综合利用像素的灰度信息和邻域空间位置信息实现有效分割.实验证明,本文算法能够更好地保留分割结果的边缘性,充分保证周围神经图像分割的精确性.

Peripheral nerve CT image segment is the basis of peripheral nerve 3D reconstruction and visualiza-tion. This paper makes an analysis of graph cut theory and algorithm, and puts forward an interactive segmenta-tion algorithm for peripheral nerve image. Based on interactive image segmentation, the algorithm uses ASIGMMin the modelling, which can cut the graph with the utilization of pixel gray and neighbourhood location informa-tion. Both theoretical analysis and simulation results justify the feasibility of the algorithm above. Peripheralnerve segmentation algorithm proposed in this paper can effectively suppress noise in the image, achieve betterretention of marginal results, and guarantee the accuracy of segmentation.