Page 329 - Fundamentals of Light Microscopy and Electronic Imaging

312 IMAGE PROCESSING FOR SCIENTIFIC PUBLICATION

• Display highly manipulated features as an inset or separate figure. Ordinarily,
cellular features should not be differentially enhanced or diminished (dodge and
burn operations in Photoshop) unless their very visibility is at stake. It is better to
show the whole cell normally, with highly manipulated features included as an
inset or as a separate picture and with processing details explained in the accom-
panying legend.

THE USE OF COLOR IN PRINTS AND IMAGE DISPLAYS

The eye is extremely sensitive to color, so assigning a color to a fluorochrome can be an
excellent way to represent two or more different signals in the same image. Because we
perceive the visual world in color and perceive color when looking in the microscope,
the presence of color makes prints and monitor displays appear natural and attractive.
Therefore, in fluorescence imaging involving multiple fluorochromes, a common
approach is to assign a monochrome color channel to each gray-scale fluorescence
image and display the information as a composite color print. Thus, color allows us to
view multiple channels simultaneously and is especially effective in demonstrating the
colocalization of two or more signals.
To preserve the dynamic range of original pixel values, color images should be han-
dled and stored in as high level a data type as is practical and feasible. This is particu-
larly important when image files are submitted directly for publication. Thus, data type
is as important for color images as it is for gray-scale image files. Previously, color was
limited to 8 to 12 bits per color channel, but because of improvements in computer speed
and storage space, recent versions of image-processing software such as Photoshop can
now support 16 bits per color channel. While monitors and printers are still limited at 8
bit resolution, color image files can be handled and stored in data types with much
higher bit depth. At the present time, however, many programs still require conversion
to byte (8 bits) for displaying and storing 24 bit color images.
Despite the advantages of using color and the improvements in handling and stor-
ing color images at high resolution, displaying multiple color signals in a single image
presents a unique set of problems:

• Brightness and visual dynamic range, the range of light intensities, and the amount
of detail that is perceived by the eye when looking at a picture (either a print or a
computer monitor) are potentially much greater in a gray-scale image than in a
color image.
• Some colors are perceived as being much brighter than others, owing to differences
in the quantum efficiency of the eye at different wavelengths (see Chapter 2), a fact
that can bias image displays.

On monitors and on prints, conversion to color can reduce brightness and visibility.
While color can increase information content in an image by including multiple chan-
nels of color-encoded information, color can also reduce the brightness and visual
dynamic range of light intensities perceived by the eye compared to the same image pre-
sented in full gray-scale range. Thus, on a monitor or on a print, color images appear
darker because they have reduced light intensities. The reduction in intensity in going