Observed versus Calculated Quadrants

I'm looking at the thon rings after CTF estimation using CTFFIND4 in RELION. The experimental thon rings match those estimated by CTFFIND4 well. However, there seems to be a streaking/blurring artifact at the intersection of all quadrants creating a cross through the image. What causes this artifact? Should these micrographs be excluded from further processing? (Micrographs collected from DED camera in movie mode)

Hello,
My first guess is that there is some artefact in the micrograph which is causing this, and that this afterfact probably doesn't necessarily mean you shouldn't use this micrograph. Have you looked carefully at the micrograph itself? Are there artefacts anywhere? Perhaps at the edges?
Another possibilty is that there is a strong density gradient across the image, such that (for example) the right edge of the image is significantly darker (say) than the left edge. Ctffind tries its best to remove these types of artefacts from micrographs, but the algorithm is not foolproof.