Detecting Electrons: CCD vs Film

1
Detecting Electrons CCD vs Film

• Practical CryoEM Course
• July 26, 2005
• Christopher Booth

2
Overview

• Basic Concepts
• Detector Quality Concepts
• How Do Detectors Work?
• Practical Evaluation Of Data Quality
• Final Practical Things To Remember

3
Basic Concepts

• Fourier Transform and Fourier Space
• Convolution
• Transfer Functions
• Point Spread Function
• Modulation Transfer Function
• Low Pass Filter

4
Fourier Transform

• The co-ordinate (?) in Fourier space is often
  referred to as spatial frequency or just
  frequency

5
Graphical Representation Of The Fourier Transform
6
Convolution
7
Convolution In Fourier Space

• Convolution in Real Space is Multiplication in
  Fourier Space
• It is a big advantage to think in Fourier Space

8
Low Pass Filter

• Reducing or removing the high frequency
  components
• Only the low frequency components are able to
  pass the filter

x

9
Transfer Functions

• A transfer function is a representation of the
  relation between the input and output of a linear
  time-invariant system
• Represented as a convolution between an input and
  a transfer function

10
Transfer Functions

• In Fourier Space this representation is
  simplified

x
11
Point Spread Function (PSF)

• The blurring of an imaginary point as it passes
  through an optical system
• Convolution of the input function with a

12
Modulation Transfer Function (MTF)

• A representation of the point spread function in
  Fourier space

x
13
Summarize Basic Concepts

• Fourier Transform and Fourier Space
• Convolution describes many real processes
• Convolution is intuitive in Fourier Space
• Transfer Functions are multiplication in Fourier
  Space
• MTF is the Fourier Transform Of the PSF
• MTF is a Transfer Function
• Some Filters are easiest to think about in
  Fourier Space

14
Detector Specific Concepts

• Nyquist Frequency
• Dynamic Range
• Linearity
• Dark Noise

15
Nyquist Frequency

• Nyquist-Shannon Sampling Theorem
• You must sample at a minimum of 2 times the
  highest frequency of the image
• This is very important when digitizing continuous
  functions such as images

16
Example Of Sampling Below Nyquist Frequency
17
Quantum Efficiency

• The Quantum Efficiency of a detector is the ratio
  of the number of photons detected to the number
  of photons incident

18
Dynamic Range

• The ratio between the smallest and largest
  possible detectable values.
• Very important for imaging diffraction patterns
  to detect weak spots and very intense spots in
  the same image

19
Linearity

• Linearity is a measure of how consistently the
  CCD responds to light over its well depth.
• For example, if a 1-second exposure to a stable
  light source produces 1000 electrons of charge,
  10 seconds should produce 10,000 electrons of
  charge

20
Summarize CCD Specific Terms

• Nyquist Frequency, must sample image at 2x the
  highest frequency you want to recover

Quantum Efficiency () Dynamic Range Linearity
CCD 50 90 10,000 Very linear
Film 5 20 100 Limited linearity
21
So Why Does Anyone Use Film?

• For High Voltage Electron Microscopes, the MTF of
  Film is in general better than that of CCD at
  high spatial frequencies.
• If you have an MTF that acts like a low pass
  filter, you may not be able to recover the high
  resolution information

22
How a CCD Detects electrons
23
Electron Path After Striking The Scintillator
100 kV
200 kV
300 kV
400 kV
24
How Readout Of the CCD Occurs
25
How Film Detects Electrons
Incident electrons
Silver Emulsion
Film
26
Silver Grain Emulsion At Various Magnification
27
How Film Is Scanned
Incident Light
Developed Silver Emulsion
Film
Scanner CCD Array
28
Options For Digitizing Film
29
Summary Of Detection Methods

• Scintillator and fiber optics introduce some
  degredation in high resolution signal in CCD
  cameras
• Film scanner optics introduce a negligible
  amount of degredation of high resolution signal

30
Practical Evaluation Of The CCD Camera
31
Decomposing Graphite Signal
x
x
32
Calculating Spectral Signal To Noise Ratio

• Signal To Noise Ratio is more meaningful if we
  think in Fourier Space

33
Calculating The Fourier Transform Of an Image
Also called the power spectrum of the image

• Image Of Carbon Film
• amorphous (non crystalline) specimen
• not beam sensitive
• common

34
Power Spectrum Of Amorphous Carbon On Film and CCD
35
Comparing The Signal To Noise Ratio From Film and
CCD
36
Film Vs CCD Head-To-Head
CCD Film
Linearity
Quantum Efficiency
Dynamic Range
MTF
37
Calculating SNR for Ice Embedded Cytoplasmic
Polyhedrosis Virus
38
Reconstruction To 9 Å Resolution
39
Confirming A 9 Å Structure
40
Relating SNR(s) To Resolution
2/5 Nyquist Frequency
41
Further Experimental Confirmation Of 2/5 Nyquist
Table 2 Comparison of Reconstruction Statistics
between Several Different Ice Embedded Single
Particles Collected On the Gatan 4kx4k CCD at 200
kV at the Indicated Nominal Magnification
Complex Number Of Particles Nominal Microscope Magnification Expected Resolution (Å) at 2/5 Nyquist Final Resolution (0.5 FSC cutoff, Å) Software Package For Reconstruction
CPV 5,000 60,000 9 9 SAVR
GroEL 8,000 80,000 6.8 7-8 EMAN
Ryr1 29,000 60,000 9 9.5 EMAN
Epsilon Phage 15,000 40,000 13.6 13 EMAN/SAVR
42
Evaluate Your Data To Estimate The Quality Of
Your Imaging

• You can use ctfit from EMAN to calculate a
  spectral signal to noise ratio
• Built In Method
• Alternate Method Presented Here

43
Final Practical Things to Remember

• Good Normalization Means Good Data
• Dark Reference
• Gain Normalization
• Quadrant Normalization
• Magnification Of CCD relative to Film
• Angstroms/Pixel

44
Normalization

• Standard Normalization
• Quadrant Normalization

45
Quadrant Normalization
46
Dark Reference
47
Gain Normalization
48
How Do I Tell If Something Is Wrong?
49
Magnification Of CCD relative to Film

• 2010F Mag x 1.38 2010F CCD Mag
• 3000SFF Mag x 1.41 3000SFF CCD Mag
• This has to be calibrated for each microscope
  detector.

50
How Do I Calculate Angstroms/Pixel?

• Å/pixel Detector Step-Size/Magnification
• For a microscope magnification of 60,000 on the
  3000SFF
• Å /pixel 150,000 Å / (microscope magnification
  x 1.41)
• Å /pixel 150,000 Å / (60,000 x 1.41)Å /pixel
  1.77

51
Conclusion

• Understand what you are trying to achieve and use
  the detector that will make your job the easiest
• Check Your Own Data!