Image processing methods in tagged magnetic resonance imaging

Tagged magnetic resonance imaging is a new method to image motion using a magnetic resonance scanner. Tags are created by changing the local magnetization of the sample prior to ordinary imaging. The magnetization is modified using the same gradient and RF coils used in imaging; thus, the tags are completely noninvasive. Because the tags are part of the sample they move when the sample moves, creating a moving pattern which can be analyzed to determine motion. Several approaches to automatically analyze the tag patterns and their motion have been proposed and developed. This paper describes two approaches, one using feature identification and tracking and another using gradient-based optical flow techniques. Recent developments in full three-dimensional motion analysis of magnetic resonance tagged images are also described: the data geometry, observation equations, regularization methods, and computational algorithms. The paper concludes with a discussion of the state-of-the-art and directions for future research.