Contrast Transfer Function (CTF)

The CTF is a sinusoidal function that oscillates between -1 and +1 as a function of spatial frequency. At specific frequencies, the CTF value crosses zero, resulting in complete loss of information at those resolutions. These zero crossings create the characteristic Thon rings visible in the power spectrum of a cryo-EM micrograph.

• Phase flipping: Negative CTF lobes invert image contrast, requiring computational sign correction.
• Frequency-dependent: The oscillation frequency increases with defocus, pushing the first zero crossing to lower resolution.
• Information gaps: Data at zero crossings cannot be recovered from a single image, necessitating data collection at multiple defocus values.

The CTF is multiplied by a damping envelope that attenuates signal at high spatial frequencies. This envelope arises from partial temporal and spatial coherence of the electron beam.

• Temporal coherence envelope: Caused by chromatic aberration and energy spread of the electron source, attenuating high-resolution information exponentially.
• Spatial coherence envelope: Arises from the finite size of the electron source and beam convergence angle, imposing a Gaussian-like damping.
• B-factor: The combined envelope is often approximated as a single Gaussian decay, parameterized by a B-factor that quantifies signal attenuation.

In the presence of objective lens astigmatism, the CTF loses its circular symmetry and becomes elliptical. Defocus varies as a function of azimuthal angle, producing an elliptical Thon ring pattern in the Fourier transform.

• Two-defocus model: Astigmatism is parameterized by two orthogonal defocus values (major and minor axes) and an azimuthal angle.
• Computational correction: Modern CTF estimation software like CTFFIND4 and Gctf fit an elliptical CTF model to the power spectrum, correcting for astigmatism during particle extraction.
• Resolution anisotropy: Uncorrected astigmatism leads to direction-dependent resolution in the final 3D reconstruction.

Computational CTF correction restores the true signal by inverting the microscope's transfer function. Two primary strategies are employed in single-particle analysis pipelines.

• Phase flipping: Multiplies Fourier components in negative CTF lobes by -1, correcting contrast inversion without amplifying noise at zero crossings.
• Wiener filtering: Applies a frequency-dependent filter that optimally balances signal restoration against noise amplification, using an estimate of the signal-to-noise ratio (SNR).
• Per-particle correction: Modern pipelines like RELION and cryoSPARC estimate and correct the CTF for each individual particle, accounting for local defocus variations across the micrograph.

Beyond defocus and astigmatism, the CTF includes phase shifts from spherical aberration (Cs) and higher-order aberrations. Spherical aberration causes electrons traveling at higher angles to be focused more strongly, introducing a frequency-dependent phase error.

• Cs correction: Modern microscopes with spherical aberration correctors (Cs-corrected) eliminate this phase error, simplifying the CTF to a pure defocus-dependent function.
• Coma and trefoil: Higher-order aberrations like axial coma and three-fold astigmatism introduce additional phase distortions that can limit resolution in uncorrected systems.
• Phase plate technology: Volta phase plates introduce a constant π/2 phase shift, converting the CTF from a sine-like to a cosine-like function and maximizing low-frequency contrast transfer.

The CTF manifests visually as Thon rings—alternating bright and dark concentric rings in the Fourier power spectrum of a cryo-EM micrograph. These rings encode the defocus, astigmatism, and envelope parameters of the imaging condition.

• Ring spacing: The spatial frequency of Thon ring oscillations increases with defocus; high-defocus images exhibit closely spaced rings.
• Automated fitting: Programs like CTFFIND4 and Gctf fit a theoretical CTF model to the rotationally averaged power spectrum, extracting defocus and astigmatism parameters.
• Resolution cutoff: The spatial frequency at which Thon rings disappear into the noise floor indicates the maximum information transfer limit of the micrograph.

The real-space counterpart of the CTF, describing how a point source is blurred by the microscope's optics. In cryo-EM, the PSF is the Fourier transform of the CTF and represents the impulse response of the imaging system.

• A perfect lens would produce an infinitesimal point; aberrations spread this into an Airy disk pattern
• The oscillating nature of the CTF in reciprocal space corresponds to ringing artifacts in the PSF
• Deconvolution of the PSF from the image is mathematically equivalent to CTF correction in Fourier space
• Understanding the PSF is critical for real-space refinement methods that directly optimize atomic models against density maps

Defocus is the intentional displacement of the specimen from the objective lens's exact focal plane, introducing phase contrast that makes biological specimens visible. The Scherzer focus is the specific defocus value that optimizes the transfer of information across the widest possible spatial frequency band.

• Underfocus (negative defocus) enhances phase contrast but introduces CTF oscillations
• The first zero-crossing of the CTF defines the point resolution of the microscope
• Modern cryo-EM collects data at multiple defoci to fill in missing information at CTF zeros
• Defocus is the primary tunable parameter controlling the shape of the CTF during data acquisition

A damping envelope that attenuates the CTF at high spatial frequencies, representing the cumulative loss of signal due to partial coherence and other information-limiting factors. The envelope function ultimately determines the information limit of the microscope.

• Temporal coherence envelope: Caused by chromatic aberration and energy spread of the electron source
• Spatial coherence envelope: Arises from finite source size and beam convergence angle
• The envelope decays as a Gaussian function, with the B-factor quantifying the rate of signal falloff
• Drift and specimen charging during exposure further dampen high-resolution information beyond the theoretical envelope

A rotationally asymmetric lens aberration where the defocus varies as a function of direction within the image plane, causing the CTF to deviate from perfect circular symmetry. Astigmatism produces elliptical Thon rings in the power spectrum.

• Characterized by a magnitude and an azimuthal angle defining the direction of maximum defocus
• Must be estimated and corrected during CTF parameter determination for accurate restoration
• Modern software like CTFFIND4 and Gctf fit a 2D astigmatic CTF model to the power spectrum
• Uncorrected astigmatism leads to anisotropic resolution in the final 3D reconstruction

Phase flipping is the simplest CTF correction method, inverting the sign of Fourier components in frequency bands where the CTF is negative. Wiener filtering is a more sophisticated approach that optimally weights each Fourier component by the signal-to-noise ratio.

• Phase flipping corrects the 180° phase shifts but does not restore amplitude modulations near CTF zeros
• Wiener filtering uses the formula: F_corrected = F_observed × CTF / (CTF² + 1/SNR)
• At CTF zero-crossings, Wiener filtering smoothly attenuates rather than dividing by zero
• Modern implementations like RELION's Bayesian approach incorporate CTF correction directly into the probabilistic reconstruction framework

Concentric oscillating rings observed in the Fourier power spectrum of a cryo-EM micrograph, representing the squared modulus of the CTF. Thon rings are the primary observable used to estimate defocus, astigmatism, and other CTF parameters.

• Named after Frank Thon, who first characterized them in 1966
• The ring spacing decreases with increasing defocus; larger defocus produces more closely spaced rings
• The outermost visible ring indicates the maximum resolution to which signal extends
• Automated CTF estimation algorithms fit a theoretical CTF² model to the rotationally averaged power spectrum
• Absence of Thon rings suggests severe drift, charging, or ice contamination