一种改进的非匀质医学图像分割算法

VARIATIONAL LEVEL SET METHOD BASED ON LOCAL AND GLOBAL INFORMATION

针对CV模型和LGDF模型分别存在不能正确分割灰度不均匀图像和对初始轮廓位置敏感的问题，提出一种改进的非匀质医学图像分割方法．利用全局灰度拟合能量对目标图像进行初步分割，再利用局部灰度信息吸引轮廓向目标边界运动并最终停止在目标边界．这个能量函数嵌入到带有正则项的水平集函数中，避免了水平集函数的重新初始化．实验结果表明，本文方法有效地解决了对初始轮廓位置敏感的问题，并且对弱边界、灰度不均匀和添加噪声的图像能够进行精确地分割．

Aiming at the question that CV model and LGDF model are difficult to get the correct segmentation results for the intensity inhomogeneity images and the segmentation results are very sensitive to the initial contours. We propose an improved based on local and global region in a variational level set formulation, which utilized the global intensity fitting energy to initialize the coniour close to the true boundaries by a preliminarily segmentation. Then the local intensity fitting energy is used to attract the contour and stop it at object boundaries. This energy is then incorporated into a variational level set formulation with a level set regularization term that avoids expensive reinitialization of the evolving level set function. Experimental results show that the method is effective for solving the sensitive problem about the position of initial contours, and robusting for segmenting weak boundary images, intensity inhomogeneity images, and noisy images.