A simple case: in 2D image processing a pipe is also called a layer. A final image may be composed of different overlapping layers, with different transparencies, and with different types of represented information.
The term layer is very two dimensional, so using it in volume (3D) visualization can be very confusing. That's why the term pipe is preferred here. It also suggests that the data is redirected through separate processing algorithms to the final result.
A 3D rendering may contain information coming from different sources, for example different channels in a Multi Channel image. Each information source goes through a separate processing pipe, and all pipes' visualization results are combined together in different ways to produce the final image. Here not only the transparency of each pipe result is important, but also the position of its components in space: if one pipe result is 2D (a Maximum Intensity Projection MIP, for example) and must be combined with other 3D pipes, where do we put it? At the front of the image, at its background, or somewhere in between? Or do we better keep information about the spatial location of each component of the 2D MIP and represent it accordingly to it? Now the decision is not as easy as in the case of a 2D layer...
Huygens Software some visualization and analysis tools (like the Surface Renderer) have graphics pipes that you can use to redirect your multi channel image data to the final rendering. A multi channel image has many information sources: each channel data can be represented separately, or not represented at all.
See the following example: a three channel image of a cell nucleus is visualized in Huygens' Object Analyzer. The DAPI image of the nucleus background (channel 2 in the image) is represented using a MIP pipe as a spatial reference, while some segmented objects of channel 1 and channel 0 are represented using two different Iso Surface pipes (one of these with red-yellow colors, the other one with bluish colors). The three active pipes are combined together in the final rendering, and each part of the MIP is placed at its correct place of the visualization volume so some parts of the iso-surfaces are below and others above it.
Object Analyzer has two surface pipes (named Primary and Secondary surface pipes) for you to put image data through for object analysis, another pipe to simultaneously visualize a MIP of one of the channels, and another pipe to visualize the ROI.
'Secondary' doesn't mean here 'less important', it is just that you can use the first pipe (Primary) and the second pipe (Secondary), abbreviated P and S in some places. We keep numbers to refer to image channels, which is something different, because in each of these pipes one can put any image channel, in any order. One can even segment the same channel twice with different conditions in each of the pipes if necessary!