基于分层的FCM算法在医学图像分割中的应用

Medical Image Segmentation Based on Layered FCM

图像分割是医学图像处理的关键问题和核心技术,医学图像中部分容积效应(PVE)、强度不一致(IIH)等现象的存在使该问题更为困难。作为有效处理PVE现象的FCM算法,在运行效率上面临着很大挑战。本文提出了一种基于分层的FCM算法,其核心思想是假设医学组织器官的聚类中心仅仅取决于隶属该组织器官的像素,而与隶属于其他组织器官的像素无关。通过这样处理,不仅可以保证较好的分割效果,从一定程度上改善强度不一致现象,同时可以提高算法的运行效率。在不同医学图像上的分割结果验证了算法的有效性与可靠性。

Image segmentation is focused and difficult in the aspect of medical image processing,and the phenomena of the partial volume effect and the intensity inhomogeneity makes the segmentation more difficult.Though the method of FCM can solve the problem of partial volume effect,the computational cost is very big.A method of medical segmentation based on layered FCM has been proposed in this paper.The premise of this method is assumed that the cluster centroids are produced from the same or similar organ or tissue,rather than from others.Then the segmentation intensity inhomogeneity can be solved,moreover the run time of the proposed method is decreased dramatically.The experimental results on different medical images verify the efficiency and effectiveness of the proposed method.