基于有偏场校正的MRI脑组织分类法

Bias Field Correction Based Tissue Classification of MR Images of Brain

核磁共振图像的自动图像分割和组织分类至今仍是一个有待解决的问题。在理想的情况下，各类组织的灰度呈正态分布；但受ＲＦ线圈、ＭＲ设备的操作环境等的影响，图像的灰度均匀性变差，相当于在增益场上叠加了一个偏移场，使信号产生混淆。作者采用“适配分割算法”，通过计算有偏场，并对图像进行灰度校正。

Automatic segmentation and tissue classification of magnetic resonance images is still problematic in applications.The distribution of the intensities of each tissue class is ideally normal.But in practice,spatial intensity inhomogeneities due to RF coils and the operating conditions of the MR equipment frequently exist which cause the distributions of intensities of different tissues to overlap significantly.The intensity inhomogeneities are modeled with a spatially varying factor called the gain field,on which an additive bias field exists.In this paper,a new method called adaptive segmenation is described.It iteratively estimates the bias field and corrects the intensity inhomogeneities.It has been proved that adaptive segmentation is an effective algorithm of segmenting normal brain tissue.