Deconvolution and Denoising for Confocal Microscopy - Modeling in Computational Biology and Biomedicine - page 145

4.2.4.4 Localizing Macromolecules

As mentioned in Sect. 4.1.2 , cell biologists are highly interested to precisely localize
macromolecules in their sub-cellular compartments. To find out if two or more
proteins possibly act in the same biological pathway, it is important to determine if
they are co-localizing. Two proteins can be said to colocalize if one protein is present
in the same voxel as the second protein. Since co-localization is of course resolution
dependent, deconvolution may be able to increase the precision of an actual protein-
protein relationship. In [ 67 ], the authors elegantly illustrates this and provides
experimental data showing that deconvolution also decreases the possibility of false
co-localization events. The authors used confocal microscopy and immunofluores-
cence labeling to determine the degree of co-localization of two receptor proteins
Dihydropyridine (DHPR) and Ryanodine (RyR) as well as to follow the changes
in their co-localization distribution during Cardiomyocytes development. This work
clearly illustrated that deconvolution of confocal images has a positive impact on
the reliability of quantitative co-localization measurements by increasing resolution
(also verified in [ 41 ]) and decreasing false co-localization events.

4.2.4.5 Improving the Resolution of STED

In [ 85 ], the authors showed that linear deconvolution [ 79 ]couldbeusedeven
in the STED microscopy technique, and revealed for the first time that nano
scale arrangement of the protein SNAP-25 is not randomly spread out on the
plasma membrane of a fixed neuronal cell, but organized in clusters of less
than 60 nm in size. This important finding could have important implications for
synaptic transmission in neurons. SNAP-25 is a part of the SNARE protein family,
which co-determines the membrane site at which a vesicle may dock and fuse.
In order to visualize the spatial distribution of this protein, it was labeled with
primary and Atto532-stained secondary antibodies. Thus, STED in combination
with deconvolution techniques not only provides sub-diffraction images but also
gives new insights into the biological data.

4.2.4.6 Improvement in Volume Estimation

In [ 26 ], the authors compare the effective improvement in the quantitative image
before and after image restoration. They measured the volume of the reference
objects using the isotropic Fakir method. They found that in each experiment
the estimates of the raw data volume were too large while measurements made
on deconvolved images matched the actual object well. For example, when 2 m

fluorescent beads where imaged in oil immersion-medium, the radius of the
specimen in the acquired image was calculated to be 2 . 28

μ

± 0 . 66 μ mandafter

deconvolution as 1 . 99 ± 0 . 49 μ m.