RESUMO
OBJECTIVES: 2D gel electrophoresis (2-DE) is the method of choice for analyzing protein expression in the field of proteomics, for example, comparing a reference with a test population. However, due to complex physical and chemical processes the locations of proteins generally vary in different 2-DE images. To cope with these variations, accurate geometric alignment of 2-DE images is important. METHODS: We introduce a new elastic registration approach for 2-DE images, which is based on an analytic solution of the Navier equation using Gaussian elastic body splines (GEBS). With this approach cross-effects in elastic deformations can be handled, which is important for the registration of 2-DE images. In addition, landmark correspondences can be included to aid the registration in regions which are difficult to register using intensity information alone. RESULTS: We have successfully applied our approach to register 2-DE gel images of different levels of complexity. In each case, gel images from a reference group are compared with a test group. To analyze the performance of our approach, we have carried out a quantitative evaluation of the registration results. Moreover, we have performed an experimental comparison with a previous elastic registration scheme. CONCLUSIONS: From the results we found that our approach is well-suited for the registration of 2-DE gel images of different levels of complexity and it turned out that the approach is superior to a previous hybrid scheme. Moreover, our approach is well-suited in a fully automatic setting and the performance can further be improved when landmark correspondences are available.
