Hi,
Contaminants or hole edges often locally affect the results of Patch Motion and Patch CTF (see example below). In datasets where there are a lot of contaminants, this can become a problem.
Would it be possible to add some kind of outlier rejection or image masking scheme so that these points are ignored during motion correction and CTF estimation?
I think the main issue with surface contaminants like these is that (1) they are very high contrast, and so dominate local patches and (2) their motion (&defocus) is likely uncorrelated with that of particles embedded in the nearby ice.
Cheers
Oli