Novel algorithm for simultaneous bias field estimation and image segmentation of MR image
一种新的核磁共振图像偏场估计和分割算法研究*

In digital medical image analysis, the intensity inhomogeneity of MR( magnetic resonance) image can seriously af-fect the performance of the algorithms, and thus the bias field has to be estimated to eliminate inhomogeneity. This paper pro-posed a novel algorithm to achieve the bias field estimation of the MR image, as well as MR image segmentation. The MMSE(minimum mean square error) criterion was applied to build the objective function. A constraint term was constructed basedon the smoothness and the local features of the bias field in order to accelerate convergence and get a better performance. Theexperimental results show that the proposed algorithm segments the image and corrects the bias field precisely. Moreover, theconstraint term can speed up the convergence and improve the performance.