改进的非局部FCM脑核磁共振图像分割与偏移场恢复耦合模型

Brain MRI Segmentation and Bias Correction Model Based on Improved FCM with Non-local Information

核磁共振图像技术可用于对疾病的辅助诊断,然而受成像机制的影响往往图像中含有噪声以及偏移场,使得传统的模糊C均值(FCM)算法很难得到较好的分割结果.为此,提出一种基于FCM算法的分割与偏移场恢复耦合模型.首先将偏移场耦合到模型中,以降低灰度不均匀对分割的影响;其次将非局部信息融入模型中,使其在降低噪声影响的同时还能保持细长拓扑结构区域信息;最后引入隶属度正则项,以降低隶属度在过渡区域的影响,改善模型的分割效果.实验结果证明,文中模型对噪声具有较好的鲁棒性,并且在分割过程中能较好地恢复图像偏移场,得到较理想的分割结果及偏移场估计.

The technology of magnetic resonance （MR） image can be used for auxiliary diagnoses of diseases. However, some image mechanisms often make images contaminated by noise or bias field, which makes the traditional fuzzy C means （FCM） algorithm difficult to obtain good segmentation results. For that, in the paper, we proposed a novel model based on FCM which combines segmentation and bias correction. Firstly, we take the bias field into the model to reduce the effect of intensity inhomogeneity; secondly, integrating the non-local information into the model can reduce the impact of noise as well as keep the image structures; finally, we introduce membership regular term to obtain crisp membership, so that the effect of membership at the transition area can be reduced, and the result of classification can be improved. Experimental results of the brain MR images show that the proposed method can reduce the impact of noise and bias field can be recovered in the process of segmentation, then obtain better segmentation results as well as the bias estimation in an accurate way.