3.4.4 Data Collection: Choosing Microscope Defocus
Defocusing the microscope’s objective lens is a convenient and conventional approach
to attaining phase contrast of the specimen by manipulating the CTF to provide contrast
at desired spatial frequencies. As a rough approximation, the higher the defocus, the
higher the low-resolution contrast, but the lower the contrast at high resolution. The
user must therefore carefully choose the optimal defocus for the biological question and
sample thickness (thicker samples will require higher defocus to compensate for lower
signal-to-noise ratio). For whole-cell tomography projects, defocus is set relatively
high (on a 300 kV microscope, typically between −5 lm and −15 lm, to provide a
first zero of *1/3 nm
−1
–*1/5 nm
−1
). Because the signal-to-noise ratio beyond the
first zero is very low, the image is usually filtered beyond the first zero, as discussed
further below. For averaging projects, defocus tends to be set closer to focus (on a
300 kV microscope, typically between −2 lm and −8 lm, which provides a first zero
around * 1/2 nm
−1 to *1/4 nm
−1
) so as to extend the resolution up to the first zero,
and reduce attenuation of signal past the first zero due to rapid CTF oscillation; this data
can subsequently be recovered by CTF correction and averaging because averaging can
boost the signal-to-noise ratio above negligible values.
3.4.5 Data Collection: Magnification
Having determined the optimal electron dose and defocus it is important to consider
the best magnification for data collection. In the digital era it is perhaps more
appropriate to think in terms of pixel size rather than magnification to set Nyquist
frequency to a suitable value. Oversampling by using pixel sizes considerably smaller
than necessary for a required Nyquist frequency is often important because DQE is
optimal at frequencies considerably lower than Nyquist frequency. If magnification is
set too high, however, the field of view will decrease—as will the electron dose per
pixel. This results in decreased signal-to-noise ratio, and fewer particles in the field of
view, necesitating collection of more data to compensate if performing subtomogram
averaging. In cameras with relatively poor MTFs, binning can be used to increase the
electron dose per pixel and improve the MTF [56]: the signal from a cluster of
adjacent pixels (typically four pixels in a 2 Â 2 block) are averaged into a single
pixel, doubling the effective pixel size, increasing the signal to noise ratio, but
decreasing spatial resolution by quartering the number of pixels.
3.4.6 Data Collection: Tilt Scheme Considerations
After configuring microscope settings for data collection, the data acquisition tilt
scheme must be determined. The number of projection images, their tilt angle
74
J. L. Ferreira et al.
Defocusing the microscope’s objective lens is a convenient and conventional approach
to attaining phase contrast of the specimen by manipulating the CTF to provide contrast
at desired spatial frequencies. As a rough approximation, the higher the defocus, the
higher the low-resolution contrast, but the lower the contrast at high resolution. The
user must therefore carefully choose the optimal defocus for the biological question and
sample thickness (thicker samples will require higher defocus to compensate for lower
signal-to-noise ratio). For whole-cell tomography projects, defocus is set relatively
high (on a 300 kV microscope, typically between −5 lm and −15 lm, to provide a
first zero of *1/3 nm
−1
–*1/5 nm
−1
). Because the signal-to-noise ratio beyond the
first zero is very low, the image is usually filtered beyond the first zero, as discussed
further below. For averaging projects, defocus tends to be set closer to focus (on a
300 kV microscope, typically between −2 lm and −8 lm, which provides a first zero
around * 1/2 nm
−1 to *1/4 nm
−1
) so as to extend the resolution up to the first zero,
and reduce attenuation of signal past the first zero due to rapid CTF oscillation; this data
can subsequently be recovered by CTF correction and averaging because averaging can
boost the signal-to-noise ratio above negligible values.
3.4.5 Data Collection: Magnification
Having determined the optimal electron dose and defocus it is important to consider
the best magnification for data collection. In the digital era it is perhaps more
appropriate to think in terms of pixel size rather than magnification to set Nyquist
frequency to a suitable value. Oversampling by using pixel sizes considerably smaller
than necessary for a required Nyquist frequency is often important because DQE is
optimal at frequencies considerably lower than Nyquist frequency. If magnification is
set too high, however, the field of view will decrease—as will the electron dose per
pixel. This results in decreased signal-to-noise ratio, and fewer particles in the field of
view, necesitating collection of more data to compensate if performing subtomogram
averaging. In cameras with relatively poor MTFs, binning can be used to increase the
electron dose per pixel and improve the MTF [56]: the signal from a cluster of
adjacent pixels (typically four pixels in a 2 Â 2 block) are averaged into a single
pixel, doubling the effective pixel size, increasing the signal to noise ratio, but
decreasing spatial resolution by quartering the number of pixels.
3.4.6 Data Collection: Tilt Scheme Considerations
After configuring microscope settings for data collection, the data acquisition tilt
scheme must be determined. The number of projection images, their tilt angle
74
J. L. Ferreira et al.