A mutual information approach to automatic detection of heart rejection in MRI
Chang H, Moura J, Wu Y, Ho C
Carnegie Mellon University
This paper develops an automatic algorithm to detect early rejection of transplanted hearts imaged by tagged MRI. We consider a mutual information based approach that compares the similarities of the regional motions between a test heart and a healthy one. Low mutual information means that the test regional motions are quite different from the healthy template and are classified as abnormal. The experiment shows good results of our algorithm.