TTFcorfirst
is a boolean switch that decides if the TTF should be corrected before or after the unbending
TTF correction before unbending (not recommended) ? [ttfcorfirst]
allows to choose between two algorithms that correct the Tilted Transfer Function (TTF) either before (yes) or after (no) unbending. The usual procedure would be to first unbend the raw image, and then calculate the FFT of the unbent image. TTBOX can then evaluate the spot amplitudes and phases and their local background under consideration of the spot-splitting due to the TTF. This is usually a sufficiently well working algorithm.
At high tilt angle, strong defocus and an excellent image with highest coherence in the electron microscope's illumination, but a bad 2D crystal, the above algorithm might risk loosing precious information: The unbending usually moves blocks of 25 pixels around in the image. If the high defocus produced Point Spread Functions (PSF) in the image that are larger than 25 pixels, then the unbending (moving) of 25-pixel blocks might cut these PSF into pieces, thereby bringing the outer fringes of the PSF out of phase with respect to the center of the PSF. This would result in loss of the high-resolution information that was encoded in the outer PSF fringes.
To avoid this, the current scripts Unbend I and Unbend II allow to correct the TTF first, and then unbend the TTF-corrected image. This is done by applying the MRC program TTMASK to the raw image first: The image is Fourier transformed, the FFT is corrected for its spot-splitting and CTF phase alterations, thereby also masking the FFT to only the information within the spots, and the result is back-transformed into real space. The result is a real-space image that is free of any CTF or TTF. However, a Fourier masking has been applied already to this image, and this happened before the unbending had taken place. The Point Spread Function (PSF) should now consist of only one pixel, instead of large fringes. If now the unbending tries to correct any crystal defects, then the PSF cannot be damaged any more. However, the signal-to-noise ratio may be lower than in the original image, making unbending harder. And the masking of the non-unbent image by TTMASK may have lost some crystal information.
Attempts by the Stahlberg laboratory to use this feature have not shown worse results, but also have not shown any better results. We are not convinced that this is a feature that should be used.