Realizing generalized registration of medical sequence image using mutual information criterion / 中国医学物理学杂志

ABSTRACT

Mutual information(MI) is a tool of measuring the statistical dependence between two random variables. MI between two images, being considered as two random fields, can be calculated to determine the degree of information redundancy between the image intensities of corresponding pixels in both images. In this paper, we utilize the MI criterion to realize the generalized registration between two neighboring slices in a medical image sequence (CT, MR et al) and consider that they match with each other if the MI between them is maximal. Experimental results confirm performance of our algorithm.