Automatic initial registration of 2-D electrophoresis gel images

Abstract

Two-dimensional gel electrophoresis is a biochemical technique for separating protein mixtures. After separation of proteins staining is performed and highlighted spots can be observed. The purpose for biochemist is to compare several gel images to reveal changes in expression of proteins. That gives information about state of cells or tissue. This information is valuable for disease diagnose, drug testing and so on. Comparison of gels is time consuming because of presence of nonlinear geometrical and spatial distortions. Automatic image comparison systems aren't fast and precision yet. There is demand for more powerful image comparison and analysis methods. In this paper new automatic initial registration algorithm for two-dimensional gel images is presented. Algorithm exploits invariant futures of gel images and employs combination of two similarity measures for control points to locate. These similarity measures are normalized cross-correlation and mutual information. Algorithm must locate several correct control points for initial registration of images to correct largest geometrical distortions.