Multifluorescence 2D Gel Imaging and Image Analysis - Difference Gel Electrophoresis (DIGE): Methods and Protocols - page 28

Fig. 3. The “traditional” matching approach is based on computer-aided matching between pairs of gel images. An overlay
view of the gel pairs allows estimating the success of the matching procedure. In case of matching ambiguities, the overlay
does not fi t properly ( left image ), and further matching has to be performed manually by the software operator. Several
tools like a “vector mode” may assist the operator's work ( centre image , some vectors are highlighted by black arrows ).
Ideally, all matching ambiguities can be solved manually ( right image ) .
with defi ned spot parameters like spot radius, spot volume, spot
contrast intensity, etc. it is possible to distinguish real protein spots
from dirt or scan artefacts. For spots in multifl uorescence 2D gel
images, several different detection algorithms are employed depend-
ing on the analysis software package used. As an example, the analysis
software DeCyder™ uses a so called “co-detection” algorithm.
Co-detection means that—for the defi nition of a possible spot—
information of all three gel images is combined, whereas other
spot detection algorithms only use the image information of the
internal protein standard image. After completion of the spot detec-
tion routine, spot matching has to be performed within all gels of
the MFA experiment. The inter-gel matching procedure is carried
out using the gel images of the pooled internal protein standard.
Although 2D analysis software uses advanced matching algorithms
(see Fig. 3 ), the software operator has to check the matching results
very closely for spot matching ambiguities. In particular, gel sections
with considerable separation inhomogeneities might cause mis-
matches that have to be corrected manually if possible. Remaining
spot matching ambiguities may cause incorrect or missing values
(see Note 5).
To overcome spot matching confl icts and to accelerate the
matching procedure, an alternative matching concept has been
developed ( 3 ) and was implemented in the Delta2D™ image analysis
software. Instead of performing spot detection on every individual
gel image, an image warping routine compensates for running
differences between different gels. After warping, corresponding
spots have equal positions in all images that were engaged in the
warping process. The warped images were fused in an artifi cial
image. Subsequently, a spot detection algorithm is applied on the
“fused” gel to create a consensus spot pattern. Based on this spot
pattern and on the warping information, spot boundaries can be
transferred from the fused image to the corresponding regions of the