Registration of dynamic MRI data and its impact on diagnostic process

This paper discusses impact of a novel registration algorithm for dynamic MRI data on diagnosis of rheumatoid arthritis. The algorithm is based on a hybrid Euclidean-Lagrangian approach. It was applied to data acquired with low and higheld MRI scanners. The scans were processed with region-of-interest based and voxel-by-voxel approaches before and after the egistration. In this paper, we demonstrate that diagnostic parameters extracted from the data before and after the registration vary dramatically, which has a crucial effect on diagnostic decision. Application of the the proposed algorithm signicantly reduces artefacts incurred due to patient motion, which permits reduction of variability of the enhancement curves, yielding more distinguishable uptake, equilibrium and wash-out phases and more precise quantitative data analysis.