Deconvolution and Denoising for Confocal Microscopy - Modeling in Computational Biology and Biomedicine - page 113

In the end of the nineteenth century, microscopists began to notice that it was
impossible to resolve features of less than half a micrometer in size. This realization
later led to the invention of an electron microscope that uses a beam of electrons,
instead of light, to create an image of the specimen. An electron microscope is
capable of much higher magnification and has much greater resolving power than
a light microscope. Modern electron microscopes can reach a resolution (resolution
can be defined as the minimal distance of separation between two point sources so
that they can be resolved as separate binary objects) of less than 0 . 1 nm, allowing
one to image individual atoms. However, successful examination of biological
samples is only possible after it has been chemically treated, to immobilize the
macromolecules, and then stained with heavy metals, to provide electron contrast
to the cell components. Moreover, the sample must be sliced very thin to allow
the electrons to penetrate into the sample. As a result, imaging living cells and
tissues with an electron microscope is simply impossible and it is mainly the
specimen preparation procedure that limits the resolving power to be at its best
around 1 nm. Light and electron microscopy have made it possible to unravel
details in the millimeter to nanometer scale for most living beings, from viruses
to bacteria, from unicellular to multicellular organisms belonging to the animal and
plant kingdom. However, imaging individual molecules within “alive” biological
samples with a precision to “nanometer spatial and millisecond time resolution”
remains a challenge.
With every decade, the microscope has improved, propelled by the technological
advancements of each epoch. While In the field of optical microscope, opticians
devised new ways of perfecting the lens system, and thereby increasing the quality
of the images obtained. Computers began to be used in microscopy primarily only as
a tool for improved image acquisition, storage, retrieval and display. It is only
in the last decade that the processing power of computers became much readily
available for image analysis and to improve the quality of acquired images. Since
then, computational methods such as deconvolution have been very successfully
applied to reduce out-of-focus light in biological samples, and to extract sharp three-
dimensional (3-D) reconstructions of the raw data from 3-D wide-field microscopy.
We will see that the progression from macro [ 61 ]to nano [ 34 ] scales, and beyond
can be partly credited to these early advancements.

4.1.2 Imaging by Fluorescence

In life sciences, living or chemically fixed cells are usually partially transparent
(optically). In general, the microscopes use properties of the sample such as absorb-
tion and refraction index variations to look at cellular structures. Nevertheless, one
of the most important methods to generate contrast in biological samples comes
from the use of fluorescent molecules. These molecules can occur naturally in
cells. Otherwise, the proteins in the sample can be labeled with specific fluorescent
proteins ( fluorophores ) or specific amino-acid moieties that can be labelled with
arsenical dyes that fluoresce under light illumination. The ability to specifically