3.4.7 Data Collection: Energy Filters and Phase Plates
Finally, if the microscope is equipped with an energy filter or phase plate, the user
should decide whether to use them. An energy filter will invariably improve image
quality; the only user-tunable parameter is filter width. In ECT the typical size of a
filter slit is around 20 eV (centered at the ‘zero loss’ energy, or incident energy of
the electron beam) as inelastic scattering events are known to remove more than
10 eVs worth of energy (typically 100–1000 eV) [65] and the temporal coherence
of current FEGs is extremely good. With a phase plate the only consideration that
the user will have is whether its use will impede throughput.
3.5 Tomogram Reconstruction: From Tilt Series Data
to Biological Insights
Having acquired one or more tilt series, the user must reconstruct a 3-D tomogram,
a process divided into the two steps of alignment and reconstruction. Many software packages exist for this, and it is usually possible to perform different stages
with different packages. Widely-used packages include IMOD [66], protomo [67],
SPIDER [68], TOM Toolbox [69], and TomoJ [70]. Dedicated programs have also
been developed to tackle individual steps such as automated fiducial alignment
[71], multicore tomographic reconstruction [72], CTF-correction [73] and
processor-intensive filtering [74].
3.5.1 Alignment
Alignment of the tilt series is the process of precisely determining the relative
orientations of each frame of the tilt series to others so as to be able to accurately
compute a 3-D tomogram, and is comprehensively covered in Chap. 7. In brief,
relative rotations and translations must be determined, as must subtle magnification
and rotation changes. To best deduce these transformations it is conventional to add
high-density gold fiducial markers for ease of tracking.
After tilt series acquisition, if images are collected on a DDD camera in movie
mode, motion correction is performed prior to any alignment. Next, tilt series
frames are coarsely aligned by cross-correlation, then fiducial markers are picked
and tracked in 2-D either manually, semi-automatically, or fully-automatically to
determine a 3-D model of fiducial locations. This 3-D model provides the transformations required to accurately align individual frames in the tilt series for
reconstruction. Alternatively, a number of fiducial-less alignment strategies can
track high-contrast landmark features instead of exogenous fiducial markers
[75, 76].
3 Electron Cryo-Tomography
77
Finally, if the microscope is equipped with an energy filter or phase plate, the user
should decide whether to use them. An energy filter will invariably improve image
quality; the only user-tunable parameter is filter width. In ECT the typical size of a
filter slit is around 20 eV (centered at the ‘zero loss’ energy, or incident energy of
the electron beam) as inelastic scattering events are known to remove more than
10 eVs worth of energy (typically 100–1000 eV) [65] and the temporal coherence
of current FEGs is extremely good. With a phase plate the only consideration that
the user will have is whether its use will impede throughput.
3.5 Tomogram Reconstruction: From Tilt Series Data
to Biological Insights
Having acquired one or more tilt series, the user must reconstruct a 3-D tomogram,
a process divided into the two steps of alignment and reconstruction. Many software packages exist for this, and it is usually possible to perform different stages
with different packages. Widely-used packages include IMOD [66], protomo [67],
SPIDER [68], TOM Toolbox [69], and TomoJ [70]. Dedicated programs have also
been developed to tackle individual steps such as automated fiducial alignment
[71], multicore tomographic reconstruction [72], CTF-correction [73] and
processor-intensive filtering [74].
3.5.1 Alignment
Alignment of the tilt series is the process of precisely determining the relative
orientations of each frame of the tilt series to others so as to be able to accurately
compute a 3-D tomogram, and is comprehensively covered in Chap. 7. In brief,
relative rotations and translations must be determined, as must subtle magnification
and rotation changes. To best deduce these transformations it is conventional to add
high-density gold fiducial markers for ease of tracking.
After tilt series acquisition, if images are collected on a DDD camera in movie
mode, motion correction is performed prior to any alignment. Next, tilt series
frames are coarsely aligned by cross-correlation, then fiducial markers are picked
and tracked in 2-D either manually, semi-automatically, or fully-automatically to
determine a 3-D model of fiducial locations. This 3-D model provides the transformations required to accurately align individual frames in the tilt series for
reconstruction. Alternatively, a number of fiducial-less alignment strategies can
track high-contrast landmark features instead of exogenous fiducial markers
[75, 76].
3 Electron Cryo-Tomography
77