Research On Gel Electrophoresis Image Deformation Analysis And Correction Algorithm

Proteomics is the forefront of biological chemistry and life science research field atpresent. Two-dimensional gel electrophoresis technology as the core of proteomic analysistechnology has played an important role. Two-dimensional gel electrophoresis technologycan separate thousands of proteins at the same time based on the difference of proteiniso-electric point and molecular weight, and separate proteins in the complex proteinmixtures in the two-dimensional plane. Then, we can obtain the two-dimensional gelelectrophoresis images by scanning the final result. In this image, the protein is portrayedas different shape, different size and different point of gray. Because of the protein pointdata quantity, gel image analysis based on computer is one of the important steps ofproteomic analysis, which mainly includes image preprocessing (noise removing, thebackground correction etc.), quantitative protein detection, and gel matching steps. Thepurpose is to accurately detect and quantify the protein characteristics. Among them, thedeformation which was carried out on the gel image distortion correction is a critical stepin image preprocessing.In gel image acquisition process, due to the different structure between proteinsamples, such as in the different salt concentration, affect of the protein migration situation,the same protein points on different gel image may appear in a different position, which isknown as the "distortion". Even if for a specific sample repeated analysis, the gel imagedistortion may happen due to the local stretch of the gel. Due to the defect of gelelectrophoresis technology itself, there always exists gel image distortion. In order toimprove the gel image matching accuracy and the analysis result, image deformationcorrection of distortion must be done. At present, the two-dimensional gel electrophoresisimage deformation algorithm can be carried out by two different approaches: spot-based and pixel-based methods.Gel image correction algorithm is studied in this thesis. This thesis elaborates thebasic knowledge of the gel image deformation correction systematically, and introduces theconcept of digital image deformation, the causes of gel image distortion, and two kinds ofgel image deformation algorithm used at home and abroad currently: spot-based methodand pixel-based method. According to the characteristics of the gel image distortion, thisthesis focuses on "smile" deformation model and thin plate spline function (TPS)deformation model. In view of the characteristics of axisymmetric in gel image distortion,the original "smiling" model is improved. On the basis of the improved "smiling"deformation model and thin plate spline function model, this thesis proposed a new methodof image deformation: First, we carry out "smiling" deformation in view of the distortioncaused by the current sliding sideways in the gel image acquisition process; Then, TPSdeformation was carried out based on the advantages of TPS deformation model ofnonlinear deformation. In comparison with the simulation of the original deformationalgorithm, the new correction algorithm can make full use of the prior knowledge of imagedistortion and the gel details, which has high calculation efficiency and is suitable forpractical application.