Deconvolution and Denoising for Confocal Microscopy - Modeling in Computational Biology and Biomedicine - page 144

effect of Poisson noise, reducing the excitation light, and as a consequence reducing
the effect of phototoxicity in living cells. The deconvolution should be considered
as an essential tool in this regard.”
In the following few paragraphs, we list some of the recent success stories from
the literature which show the usefulness of deconvolution to biology.

4.2.4.1 Contrast Improvement for Imaging Plant Cells

Moreno et al. [ 51 ] described in an elegant review on “Imaging Plant Cells” that
with particular biological samples, only deconvolution could reveal structures not
observable using CLSM and multi-photon microscopy. The authors made use
of the auto-fluorescence of a unicellular green algae to create an image of the
ultrastructure of this organism. Both, confocal and multi-photon microscopy failed
to produce sharp images taken over the entire cell depth. The main problem was that
the high intensity illumination needed for confocal imaging caused signal bleaching
as well as an emission shift from red to green [ 20 ]. Since wide-field imaging
demands much less excitation power, deconvolution of the obtained wide-field data
stack generated the contrast improvement needed to faithfully observe cell features.

4.2.4.2 Quantitative Analysis

In [ 78 ], the authors described that they could not resolve microtubular cytoskeleton
components labeled with yellow fluorescent proteins fused to tubulins in the
parasite. By contrast, when they deconvolved the image obtained from the wide-field
microscope, they were able to visualize weak fluorescent signals from individual
microtubules of the cytoskeleton. This allows the quantitative studies of very dim
fluorescent structures in these thin living cells.

4.2.4.3 Improvement of Signal-to-Noise Ratio

The use of iterative deconvolution algorithms in live-cell imaging was also demon-
strated by Platani et al. [ 59 ] for exploring the dynamics of the Cajal body within the
nucleus of living human cells. They used a HeLa derived cell line stably expressing
a fluorescent coilin protein to specifically label Cajal bodies. This line showed
to be highly photosensitive, which required the use of short exposure times and
imaging conditions with very low light levels. Once the SNR was improved, the
detection of these tiny Cajal bodies became easy, because it permitted an automated
segmentation and tracking algorithm to characterize their mobility in living cells.
They also demonstrated that iterative deconvolution algorithms when applied in life
cell microscopy proved to be essential to better visualize and explore the dynamics
of these bodies within the nuclei of living human cells.