Non-rigid biomedical image registration using graph cuts with a novel data term

Problem of non-rigid registration has become very important in the area of biomedical imaging. A non-rigid registration problem is modeled as an optimization problem and is solved using graph cuts and MRFs in recent years. In this paper, we have improved the graph cuts-based solution to non-rigid registration with a novel data term. The proposed data term has several advantages. Firstly, displacement labels can be directly assigned from this data term. Secondly, our data term imposes stricter penalty for intensity mismatches and hence yields higher registration accuracy. Finally, this data term can efficiently handle the dissimilarities in the intensity patterns between the floating and the reference images which may also arise due to some changes in illumination in addition to motion. We show the effectiveness of the proposed method on MRI images of brain and light microscopy images of retina. Experimental results indicate the superiority of our technique over some well-known non-rigid registration algorithms.