There are many considerations to be made prior to embarking on EM in general,
but even more so the volume SEM techniques. Both SBEM and FIB-SEM have
their own limitations and rely heavily on the sample preparation. The sample must
produce sufficient signal to allow imaging in backscatter mode with low accelerating voltages and withstand the effects of the electron beam to allow imaging and
sectioning/milling. These affects are amplified in SBEM, where charging is an
especially large hurdle. The interplay between resolution and imaging size is
another consideration in FIB-SEM and SBEM where there is a need to balance
imaging conditions with the mechanism to remove sections. Optimising the data
quality requires strict assessment of what is to be achieved from the data. SBEM
and FIB-SEM while overlapping in their applicability do have their own niches and
can be and used as complimentary techniques. It is not uncommon for laboratories
to have both systems and to direct to either depending on the needs of the projects.
CLEM has been utilised by both SBEM and FIB-SEM to facilitate targeting the
ROI and optimise the acquired volume [9, 37, 75, 79]. Recently Brama et al. [82]
have designed an integrated LM/SBEM in which a miniature fluorescence light
microscope is built into the stage of the SBEM allowing sequential imaging of the
same region using both modalities. At present the instrument they have built, the
miniLM, is just a proof of principle as the resolution in the images needs to be
improved.
There have been many applications of the FIB-SEM. Schertel et al. [83] were
able to apply a typical mill and image approach in a cryo-FIB-SEM on natively
frozen mouse optic nerve and Bacillus subtilis spores. Recently, cryo-FIB-SEM is
also receiving attention for its ability to prepare thin lamella for cryo-electron
tomography. This has been successful in not only cells [84, 85] but also nematodes
[71]. The benefit here is that thicker sections can be prepared (500 nm) with less
artefacts as compared to those achieved by cryo-electron microscopy of vitreous
sections (CEMOVIS), 100 nm [65].
With tools that can capture data in an automated way, the bottleneck becomes
handling this 3D volume once it is collected [86]. The datasets can be enormous
making the handling and analysis a major undertaking. Even data movement from
the instrument can be a difficult exercise. In saying this, the data collected is simply
a 3D volume and there are many softwares available for its analysis. The volume is
made up of a series of 2D images that first need to be aligned before analysis. This
is a simple problem that can be taken care of by IMOD, plugins within FIJI (for
example TrakEM2 [87], BigWarp http://fiji.sc/BigWarp; harnessing the
BigDataViewer system [88], StackReg [89]), or in house custom made scripts.
Segmentation is the best method for 3D visualisation of specific structures within
the context of the volume and can be performed with software such as Amira and
IMOD, which is a time consuming process. Programs such as Ilastik [90, 91] and
MIB [92] house tools to help with the segmentation in an attempt to make it
semi-automated.
The broad application of SBEM and FIB-SEM pushes forward the methods
alongside technological advances. Moreover, there are continual improvements to
instruments and detectors. An example of this is the multi-beam SEM with 61 [93],
142
R. I. Webb and N. L. Schieber
but even more so the volume SEM techniques. Both SBEM and FIB-SEM have
their own limitations and rely heavily on the sample preparation. The sample must
produce sufficient signal to allow imaging in backscatter mode with low accelerating voltages and withstand the effects of the electron beam to allow imaging and
sectioning/milling. These affects are amplified in SBEM, where charging is an
especially large hurdle. The interplay between resolution and imaging size is
another consideration in FIB-SEM and SBEM where there is a need to balance
imaging conditions with the mechanism to remove sections. Optimising the data
quality requires strict assessment of what is to be achieved from the data. SBEM
and FIB-SEM while overlapping in their applicability do have their own niches and
can be and used as complimentary techniques. It is not uncommon for laboratories
to have both systems and to direct to either depending on the needs of the projects.
CLEM has been utilised by both SBEM and FIB-SEM to facilitate targeting the
ROI and optimise the acquired volume [9, 37, 75, 79]. Recently Brama et al. [82]
have designed an integrated LM/SBEM in which a miniature fluorescence light
microscope is built into the stage of the SBEM allowing sequential imaging of the
same region using both modalities. At present the instrument they have built, the
miniLM, is just a proof of principle as the resolution in the images needs to be
improved.
There have been many applications of the FIB-SEM. Schertel et al. [83] were
able to apply a typical mill and image approach in a cryo-FIB-SEM on natively
frozen mouse optic nerve and Bacillus subtilis spores. Recently, cryo-FIB-SEM is
also receiving attention for its ability to prepare thin lamella for cryo-electron
tomography. This has been successful in not only cells [84, 85] but also nematodes
[71]. The benefit here is that thicker sections can be prepared (500 nm) with less
artefacts as compared to those achieved by cryo-electron microscopy of vitreous
sections (CEMOVIS), 100 nm [65].
With tools that can capture data in an automated way, the bottleneck becomes
handling this 3D volume once it is collected [86]. The datasets can be enormous
making the handling and analysis a major undertaking. Even data movement from
the instrument can be a difficult exercise. In saying this, the data collected is simply
a 3D volume and there are many softwares available for its analysis. The volume is
made up of a series of 2D images that first need to be aligned before analysis. This
is a simple problem that can be taken care of by IMOD, plugins within FIJI (for
example TrakEM2 [87], BigWarp http://fiji.sc/BigWarp; harnessing the
BigDataViewer system [88], StackReg [89]), or in house custom made scripts.
Segmentation is the best method for 3D visualisation of specific structures within
the context of the volume and can be performed with software such as Amira and
IMOD, which is a time consuming process. Programs such as Ilastik [90, 91] and
MIB [92] house tools to help with the segmentation in an attempt to make it
semi-automated.
The broad application of SBEM and FIB-SEM pushes forward the methods
alongside technological advances. Moreover, there are continual improvements to
instruments and detectors. An example of this is the multi-beam SEM with 61 [93],
142
R. I. Webb and N. L. Schieber