Elastic Image Registration with Applications to Proteomics

In proteomics, electrophoretic 2-D gels are used to separate the proteome according to the molecular weight and electrical charge of its constituents, which are proteins expressed by a cell or organ at a particular time and under specific conditions. One of the main applications is the analysis of differential expressions between different conditions for which certain (perhaps many) spots are present in one of the images, but not in the other. One of the difficulties of this analysis is that 2-D gels are affected by spatial distortions due to run-time differences and dye-front deformations, which results in images that are significantly dissimilar and that pose a challenging problem to image-registration algorithms. In this paper, we test the efficiency of a stateof-the-art elastic-registration algorithm that we had already introduced in the context of biomedical images [1]. We study here the registration of simulated 2-D gels with known expression patterns and deformations. We show that our algorithm is capable of handling such situations. The proposed algorithm is publicly available at http://bigwww.epfl.ch under the name UnWarpJ.