Study On Intensity Inhomogeneity Correction In MR Images

Magnetic Resonance Imaging becomes more and more popular in the clinical diagnosis and application, due to its multimodality imaging and safety. Due to the nonuniformity of the magnetic field and the receive coils, the intensity of the MR images is always inhomogeneous, which would affect the diagnosis and automatical analyzation and process.After summarizing some methods of intensity inhomogeneity correction, we proposed two improved methods in this paper. One is the intensity inhomogeneity correction method based on similar points. The other one is a method combining the Split Bregman iteration and the weighted CLIC model.In the intensity inhomogeneity correction method based on the similar points, similar points are selected by a method used in NLM algorithm, then a smooth polynomial surface is fitted by a modified least squares method. The method has the same time-efficient as other surface fitting methods in reconstruction. Besides, this method overcomes the limitation of previous traditional methods that require all the points selected from the same tissue.The method based on Split Bregman iteration and the weighted CLIC method introduces the information of edges in an image to CLIC method, with a larger Gaussian kernel used to preserve the smoothness of the bias and the Split Bregman iteration to speed up the process. This improved method overcome the loss of detail information occurring in the CLIC method.