Registration of Time-Series Contrast Enhanced Magnetic Resonance Images for Renography

Renovascular disease is an important cause of hypertension. For assessing treatment options for renovascular disease such as angioplasty or nephrectomy, it is important to characterize the renal tissue. Magnetic resonance (MR) renography is becoming a viable method for characterization of the renal tissue. However, analysis of MR renography is hampered by tissue motion. We investigate two automated image registration methods for minimization of the effects of tissue motion The first is semi-automated registration using contours. The second is an adaptation of the Automated Image Registration (AIR) algorithm that accommodates large-scale motion and tissue enhancement from a contrast agent. We compared the results of these methods with manual registration using image overlays. Semiautomated registration using contours accurately registered a 2D MR renography data set of 140 time frames with obvious errors in only 7 slices. With correction in those slices, semiautomatic registration had equivalent quality to manual registration. The adaptation of the AIR algorithm produced better results on 3D MR renography in healthy kidneys than manual registration but worse results in a diseased kidney. We conclude that automated registration of 2D and 3D MR renography is feasible. 1. Background Registration of time-series images acquired over the course of contrast enhancement, or socalled dynamic MR, is a challenging problem. Typically such images are acquired in tissues where large-scale tissue displacement can exist between images due to variation in inspiration for breath-hold images or, for example, due to stretching of unsupported breast tissue over the course of the exam. Furthermore, the displacement of soft tissue from one time frame to another may be complex, having both rigid and elastic components. Registration of dynamic MR is also challenging because the appearance of the tissue changes markedly over the course of the enhancement. Since the various tissue types do not enhance uniformly, the relative intensities between tissues vary with time. Several methods have been developed for registration of dynamic MR or dynamic radionuclide imaging. A landmark-based method has been developed for 2D dynamic MR. In this method a contour is drawn around the kidney in one time frame. That contour is then registered to all edge images derived from all other images in the time series. Problems related to variation in image contrast are minimized when using contours and edge images. Our 2D registration method is an extension of this method. A registration method has also been proposed for analysis of cardiac dynamic MR. In this method, a single mask is manually drawn around the 2D short-axis cross-section of the heart that most nearly approximates the heart shape without excluding the heart from any time frame. The images within the masked region for all time frames are registered to one another. The reference or standard image is the accumulated average of the registered images. The registration was performed by the least-squares criterion. The registration method used a 3parameter rigid model of motion. Our 3D registration method is an extension of this method. A registration method has also been developed for mammographic dynamic MR. The cost function in this method is the mutual information. Mutual information is a measure of similarity between images that is insensitive to differences in relative contrast between the images. This method requires only that each individual tissue within the image vary in intensity consistently from one image to another. However, no comparison is made between registration using the mutual information cost-function and, for example, a more conventional least-squares cost function. The Automated Image Registration (AIR) algorithm has been applied extensively to functional brain MR. In that application, motion between images is important but small-scale. The direct application of this algorithm has thus far not been established. We discuss a multiple-step procedure that incorporates the AIR algorithm for registration of dynamic MR of the kidneys.