Investigation of Segmentation Methods for Proteomics

Abstract

Two-dimensional electrophoresis is the most common method in the biochemistry. It is used to investigater proteom – all proteins existing in the cell or an organism at a particular time instance. The investigation results get sense when several images of twodimensional electrophoresis gels are compared. Due to huge amount of spots in the images of gels and big variability in the images under the comparison, various automatic comparison and analysis systems are built. Image segmentation is the one of the processes used during automatic two-dimensional electrophoresis gel image analysis. Quality of the segmentation is crucial for the comparison results. In this work: segmentation process and its place in the automatic gel analysis are discussed, common methods for the segmentation of images taken from two-dimensional electrophoresis gels are analyzed, alternative segmentation algorithm and results of its application are presented. Proposed segmentation algorithm consists of noise reduction based on the median filtering, contrast enhancement based on the morphological operators, and two level segmentation employing Watershed transform. Ill. 4, bibl. 19 (in Lithuanian; summaries in Lithuanian, English, Russian).