Last week’s Nature contained a fascinating article on using fluorescent proteins in transgenic mice to produce images of neurons in the brain:

Transgenic strategies for combinatorial expression of fluorescent proteins in the nervous system

Some images from the project:

^{The Center for Brain Science, Harvard University; Science Friday, NPR (used without permission)}

The authors were able to express multiple fluorescent proteins (XFPs) from a single transgene in order to label individual cells with 90 distinguishable colors. The authors point out that this could be used for detailed circuit analysis, but that the usefuless of this in analyzing complex connectivity is dependent on the number of distinguishable colors.

In the Proceedings of IEEE Visualization 1996, Chris Healy looked at what factors allow two colors to be rapidly distinguished: Choosing Effective Colours for Data Visualization. Healy found that in order to provide good differentiation between individual data elements, one can only display approximately seven colors at once and that these colors need a minimum color distance, linear separation, and color category differentiation. Selection of colors based on these criteria allows for during rapid and accurate visual searching of a display. So if one is attempting to quickly scan brain images (perhaps looking for circuits), one would probably need to reduce the number of colors used in the image, perhaps through post-processing.

I especially liked Figure 4 ,which shows the colors obtained from 3 XFP combinations, oculomotor axons (also shown in the first image above), brainstem axons, and cells in the dentate gyrus.