BMVC 2004, Kingston, 7th-9th Sept, 2004
2 Dimensional Electrophoresis Gel Registration Using Point Matching and Local Image-Based Refinement
M. Rogers, J. Graham (University of Manchester) and R.P. Tong (AstraZeneca)
Recently, proteomics research has become a large growth area in the biosciences.
Studies involve differential analysis of large sets of 2-D Electrophoresis
(2-DE) gels. We present a robust and accurate 2-DE gel alignment
algorithm which combines point matching and local image-based re-
finement. The algorithm uses a novel combination of Euclidian, shape context,
image and feature based attributes to produce a point distance measure.
Correspondence is determined using this measure and is further improved
using an iterative M-estimation approach, and shown to be robust in the presence
of large image distortions. Local image-based refinement is shown to
improve significantly alignment accuracy. The high accuracy and robustness
of the resulting system indicates that it is a promising method for use in practical
gel alignment situations.