An improved image registration and bias field correction coupled method for Brain MR images

Bias field causes considerable difficulty in the quantitative analysis of brain magnetic resonance (MR) images. Image registration is a necessary pre-processing step before quantitative analysis of brain MR data. A novel variational optical flow approach for image registration is proposed in this paper. The advantages of our method are as follows. First, we coupled bias correction and optical flow image registration within the unified variational framework in order to reduce the effect of bias field. Second, the regularization term based on the non-local information is proposed, which is able to preserve the image structure feature during the registration process. Moreover, we could recover the corrected target image through the estimated bias field. Experiments on synthetic and real brain images demonstrate the advantages of our method over Horn and Schunck(HS) model in terms of both efficiency and accuracy.