reducing radiation damage, but tests revealed that liquid nitrogen actually provides
superior results [36].
Electron microscopes perform analogously to visible light microscopes: a radiation source is projected through a specimen, and scattered radiation is focused by
lenses to form an image. The electron microscope uses electrons for illumination,
and the more coherent the beam is, the more coherent the scattered electrons that are
refocused to form a higher resolution image [7]. Contemporary electron microscopes use tungsten filaments, lanthanum hexaboride crystals, or field emission
guns (FEGs). The latter two sources are sufficiently coherent for ECT; however
when aiming for a high resolution subtomogram average, a FEG is the best source,
and will always be fitted to 300kV microscopes which are optimal for ECT. The
electrons can be accelerated from the electron gun with various energies, which
dictate the mean penetration depth before the first scattering event. Appropriate
settings are discussed below.
Image data can be recorded in a number of ways: film, charge coupled detectors
(CCDs), or direct detection devices (DDDs). Performance of these three detector
types is assessed in terms of the Modulation Transfer Function (MTF) which
describes the attenuation of spatial frequency approaching Nyquist frequency as a
result of pixel cross-talk; and the Detective Quantum Efficiency (DQE) which
describes the fraction of input signal captured by the detector. A camera with a
DQE value half that of another camera would require double the electron dose to
achieve an image of the same quality. Although film allows for large images to be
recorded at relatively high resolutions, the time required to repeatedly load and
develop it is a huge bottleneck for cryo-tomographic data collection, rendering it
unfeasible for high-throughput imaging [37]. CCD cameras detect photons produced by electrons hitting the detector and offer a large improvement in output time
as compared to film. However, although CCDs are efficient in making data
acquisition high-throughput, the random interactions of the electrons with the
scintillator comes with a resolution cost and near Nyquist frequency the performance of CCDs is lower than that of film in terms of both DQE and MTF [37].
The recent development of DDDs based on Complementary Metal–Oxide–
Semiconductor (CMOS) technology has revolutionized the field of cryo-microscopy
because by directly detecting electrons, a considerably superior MTF is achieved.
Some DDDs have a number of additional critical innovations that further their
abilities. One of the major resolution limitations in cryo-microscopy has been
specimen movement due to instability of the specimen holder, exposure to the
electron beam, or specimen charging that deflects electrons. New DDDs can be used
to correct for the blurring caused by this movement by acquiring many ‘subframes’
with sufficient signal for subsequent subframe alignment and averaging, producing a
sharper composite image. In tomography this is particularly important at high tilts
(particularly for relatively unstable specimen cryoholders) where the specimen is
thickest, giving an improved signal-to-noise ratio and providing higher resolution
images. Another major DDD innovation is to use a superresolution approach to
locate where individual electrons hit the camera by determining the peak’s centroid
to subpixel accuracy—effectively halving the pixel size and quadrupling number of
3 Electron Cryo-Tomography
69
superior results [36].
Electron microscopes perform analogously to visible light microscopes: a radiation source is projected through a specimen, and scattered radiation is focused by
lenses to form an image. The electron microscope uses electrons for illumination,
and the more coherent the beam is, the more coherent the scattered electrons that are
refocused to form a higher resolution image [7]. Contemporary electron microscopes use tungsten filaments, lanthanum hexaboride crystals, or field emission
guns (FEGs). The latter two sources are sufficiently coherent for ECT; however
when aiming for a high resolution subtomogram average, a FEG is the best source,
and will always be fitted to 300kV microscopes which are optimal for ECT. The
electrons can be accelerated from the electron gun with various energies, which
dictate the mean penetration depth before the first scattering event. Appropriate
settings are discussed below.
Image data can be recorded in a number of ways: film, charge coupled detectors
(CCDs), or direct detection devices (DDDs). Performance of these three detector
types is assessed in terms of the Modulation Transfer Function (MTF) which
describes the attenuation of spatial frequency approaching Nyquist frequency as a
result of pixel cross-talk; and the Detective Quantum Efficiency (DQE) which
describes the fraction of input signal captured by the detector. A camera with a
DQE value half that of another camera would require double the electron dose to
achieve an image of the same quality. Although film allows for large images to be
recorded at relatively high resolutions, the time required to repeatedly load and
develop it is a huge bottleneck for cryo-tomographic data collection, rendering it
unfeasible for high-throughput imaging [37]. CCD cameras detect photons produced by electrons hitting the detector and offer a large improvement in output time
as compared to film. However, although CCDs are efficient in making data
acquisition high-throughput, the random interactions of the electrons with the
scintillator comes with a resolution cost and near Nyquist frequency the performance of CCDs is lower than that of film in terms of both DQE and MTF [37].
The recent development of DDDs based on Complementary Metal–Oxide–
Semiconductor (CMOS) technology has revolutionized the field of cryo-microscopy
because by directly detecting electrons, a considerably superior MTF is achieved.
Some DDDs have a number of additional critical innovations that further their
abilities. One of the major resolution limitations in cryo-microscopy has been
specimen movement due to instability of the specimen holder, exposure to the
electron beam, or specimen charging that deflects electrons. New DDDs can be used
to correct for the blurring caused by this movement by acquiring many ‘subframes’
with sufficient signal for subsequent subframe alignment and averaging, producing a
sharper composite image. In tomography this is particularly important at high tilts
(particularly for relatively unstable specimen cryoholders) where the specimen is
thickest, giving an improved signal-to-noise ratio and providing higher resolution
images. Another major DDD innovation is to use a superresolution approach to
locate where individual electrons hit the camera by determining the peak’s centroid
to subpixel accuracy—effectively halving the pixel size and quadrupling number of
3 Electron Cryo-Tomography
69