correcting the image
play

Correcting the Image N. Grigorieff What is Wrong with the Image? - PowerPoint PPT Presentation

Correcting the Image N. Grigorieff What is Wrong with the Image? Xing Zhang 2007 Electron Diffraction 4 3.4 Baldwin & Henderson 1984 Image Contrast [ ] ( ) ( ( ) ( ) ) ( ) ( ) ( ) 2 MTF = + 2 k k k k k k P


  1. Correcting the Image N. Grigorieff

  2. What is “Wrong” with the Image? Xing Zhang 2007

  3. Electron Diffraction 4 Å 3.4 Å Baldwin & Henderson 1984

  4. Image Contrast [ ] ( ) ( ( ) ( ) ) ( ) ( ) ( ) 2 MTF = + 2 k k k k k k P S B ENV CTF i i ( ) ( ) S = signal + + 2 2 2 k k B MTF B B i = imaged background s e ENV i = CTF envelope MTF = Detector MTF B s = shot noise B e = detector noise Zeng et al. 2007

  5. Things to Worry About • Quality – Resolution, contrast • Contrast transfer function – Defocus, astigmatism, beam tilt, envelope • Magnification – Variations from image to image/within images • Ewald sphere (for large objects) • Contrast normalization • Inelastic and multiple scattering – Energy filter, diffraction theory

  6. Rotavirus Double-Layer Particles 1000 Å

  7. Side Chains Rotavirus double-layer particles 8400 particles Xing Zhang Ethan Settembre 2007

  8. Image Quality - Grids

  9. Image Quality – Location on Grid

  10. Image Quality - Defocus

  11. 3.8 Å 6.5 Å Thin Rings

  12. Determining the CTF CTFFIND3 Image power spectrum Model Experiment E = 120 kV, ∆ f = 21000 Å, C s = 2 mm, A = 0.15 Mindell & Grigorieff 2003

  13. Defocus Gradient Defocus [Å] y x CTFTILT [ μ m] van Heel et al. 2000 Mindell & Grigorieff 2003

  14. Displaced Information Δ f Θ Δ r = Δ f Θ

  15. Margins Δ = Δ Θ r f = × 25000 Å 5.3mrad = 130 Å 160 Å

  16. Ewald Sphere k L k R A B F L F R k 0 A' O B' B* A * Phase error = 60° for 700 Å particle at 300 kV

  17. Normalization ( ) ( ) − , avg ( ) I x y I ˆ = , I x y ( ) var I [ ] ( ) ( ) − , avg bg ( ) I x y I ˆ = , I x y [ ] ( ) var bg I Sorzano et al. 2004

  18. Maximum Likelihood ( ) ( ) ∫ φ φ Θ φ ( ) , n d X p 1 N ∑ + = Structure for i i ( 1 ) n ( ) A ∫ φ Θ φ ( ) n +1 iteration , n d N p = 1 i i ( ) ⎡ ⎤ 2 φ − ⎛ ⎞ M 1 ( ) ( ) X A Probability φ Θ = − φ Θ ⎜ ⎟ ⎢ ⎥ , exp i | p f σ π σ 2 i function ⎝ ⎠ ⎢ 2 ⎥ 2 ⎣ ⎦ φ : alignment parameters X i : i th image N : # of images Θ : model parameters σ : noise in images : positional probab. f Sigworth 1998

  19. Noise Model Zeng et al. 2007

  20. ML processing of 2D crystals Crystallography Alignment of individual unit cells using ML approach

  21. Systematic Error 70 60 Model phase error [deg] 50 40 Refined 3D structure 30 of bacteriorhodopsin 20 10 10 20 30 40 50 60 70 Data phase error [deg] Grigorieff et al. 1996

  22. Inelastic Image Unfiltered 0 eV 25 eV 300 kV, 6 μ m underfocus, 15 eV energy window Chen Xu 2007

  23. Things to Worry About ? • Quality – Resolution, contrast • Contrast transfer function – Defocus, astigmatism, beam tilt, envelope • Magnification – Variations from image to image/within images • Ewald sphere (for large objects) • Contrast normalization • Inelastic and multiple scattering – Energy filter, diffraction theory

  24. Acknowledgements • Rotavirus DLP Xing Zhang, Ethan Settembre (Harrison lab) • Purple membrane Fritz Zemlin, Erich Beckmann • ML of 2D crystals Xiangyan Zeng (Stahlberg lab) • CTFFIND3/TILT Joe Mindell • Ewald sphere Matthias Wolf, David DeRosier • Inelastic scattering Chen Xu • Financial Support: HHMI, NIH, NSF Humboldt Foundation

Recommend


More recommend