Visualization in Connectomics - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 215

of cells and to come to conclusions about their function. Following the motto “the
gain in the brain lies mainly in the stain” [ 1 ], the three following main techniques
are employed to map neuronal circuits with light microscopy [ 60 ].

Single-cell staining by dye impregnation. This is the oldest staining method and
it laid the foundation for modern neuroscience. As neuronal tissue is densely packed
with cells, a complete staining of the whole sample would not allow one to dis-
criminate single cells in light microscopy images. Instead, the so-called Golgi stain
enables stochastic marking of just a few individual nerve cells. The stained cells
appear dark in the light microscopy images, discriminating them from a bright back-
ground formed by the unstained tissue. This staining method, combined with the
ability of the light microscope to focus on different depth of the sample, allows for
3D imaging of the cell geometry. The famous neuroscientist Cajal (1852-1934) was
able to identify different types of neurons and also describe connectivity patterns and
principles of neuronal circuit organization using Golgi's method [ 60 ].

Diffusion or transport staining. Diffusion staining techniques enable biologists
to analyze the projective trajectory of brain regions. For this technique, different
staining markers are injected into different regions of the brain in vivo . The staining
is then diffused along the connected neurons. Finally, a sample of brain tissue is
extracted from a different region, in which no marker has been injected. The color
code in the staining of different neurons in this area then reveals the projection of
these neurons back to the initial staining areas, providing information about long-
distance connectivity [ 33 ]. The range of possible colors for this method is limited to
three or four different stainings.
Multicolor or brainbow. This staining technique does not involve application or
injection of staining to brain tissue. Instead, transgenic mice are bred to produce
photophysical fluorescent proteins. A confocal laser-scanning microscope activates
the fluorescent proteins with a laser beam and records an image with the expressed
light. Brainbowmice are bred to express three fluorescent proteins of different colors.
By different stochastic expression of these three colors, the single neurons of the mice
are colored with one out of 100 labels. The main advantage of this method is that

it allows one to uniquely identify dendrites and axons belonging to the same neuron
in densely colored tissue [ 60 ],seealsoFig. 21.3 .

>

All of these three staining methods allow imaging the geometry of neurons at
the micrometer scale. The different staining protocols all aim at visually separating
single neurons out of the complex and dense neuronal tissue. Visualization tech-
niques for connectomics need to enhance the visual separation further, e.g., by
providing contrast enhancement and enabling flexible mappings of image data to
varying amounts of transparency in the transfer function [ 51 ]. Especially for the
brainbow staining it is useful to have visual enhancement of color differences in
regions of interest where two neurons with a similar staining combination need to
be distinguished. For diffusion staining this problem is less pronounced than for
brainbow data, as typically only three to four easily distinguishable colors are used.
But this also leads to the challenge of distinguishing two neighboring cells that are