How is the background of spots calculated?
Image background is generated by material that is stained but not part of a protein spot. The method of background subtraction can have a significant influence on spot quantities; therefore, it is important that background quantities are made explicit by the software instead of being silently subtracted from a spot’s quantity. Background levels can vary considerably between regions on a gel and between gels. Some background subtraction methods are based on the gray levels at the spot boundaries, other approaches are based on the entire available image data. One example for a background estimation that is based on the spot boundary is DeCyder’s (GE Healthcare) rule: Background is determined by the tenth percentile value of all intensity values on the boundary. A background model based on local minima was used by Tyson et al. (1986). The Melanie II software (Appel et al. 1997) calculated background based on a polynomial that is fitted to image intensities. A related approach is the rolling ball method (Skolnick 1986) that determines background levels by fitting a sphere into the 3-D “landscape” of the image (see Fig. 1). The sphere needs to be large enough such that the ball will not go too deep into the spots. Background levels are then determined relative to the center of the ball when it touches the image surface.
Fig. 1: Background subtraction using the rolling ball approach
From: Berth, M., Moser, F., Kolbe, M., & Bernhardt, J. (2007). The state of the art in the analysis of two dimensional gel electrophoresis images. Applied Microbiology and Biotechnology, 76(6), pp.1223–1243.