基于FCM聚类算法的MRI脑组织图像分割方法比较研究

目的磁共振成像(magnetic resonance imaging,MRI)对脑组织有较好的成像效果,但噪声、偏移场和部分容积效应(partial volume effect,PVE)的存在,使得全自动分割MRI图像面临一定的困难。模糊C均值(fuzzy C-means,FCM)聚类算法在脑组织分割中得到较广泛研究。本文以存在噪声和偏移场影响的脑MRI图像分割为应用背景,研究了大量相关方法,探讨FCM算法分割脑部图像的改进思想。方法本文主要研究了9种FCM算法的理论基础,并通过脑组织分割实验对各种算法进行了分析。结果比较了不同算法的优劣,给出各类算法直观及定量评价结果。结论偏移场和噪声对脑磁共振图像组织分类质量有明显影响。其中几种方法可以减弱这些不利影响,但由于难以选择合适的参数,其分类效果并不理想。如何合理利用空间信息在未来仍有较大研究价值。

A comparative study for MRI brain image segmentation based on FCM clustering algorithm

Objective Magnetic resonance imaging ( MRI) provides high resolution for brain tissue , yet the presence of noise, bias field, and partial volume effect (PVE), make automatic segmentation of MRI image a challenge task .Fuzzy C-means ( FCM ) clustering algorithm is widely studied these years .This paper investigates different variants of FCM methods for brain tissue segmentation and explores its improvement , especially in the presence of noise and bias field in MRI images .Methods Nine variants of FCM methods are analyzed theoretically first .Then brain tissue segmentation experiments are done to evaluate these algorithms ’ performance .Results We compare the merits of different algorithms and give the qualitative and quantitative results . Conclusions Bias field and noise degrade the classification quality apparently .Though certain methods have the abilities to decrease the influence of noise and bias field, the difficulty of choosing the optimum parameters hinders their performance.Reasonable utilization of spatial information has research value in the future.