Noise pixels in 2D classes

During 2D classification of a negative stain dataset, some of the classes have a strong pixel with circular ripples around it that cover most of the image:


Is there a way to mitigate this during classification or maybe at earlier stages? Is it possible that the particles in those classes are good, but get misclassified due to a strong pixel?

Quite often these are due to hot pixels that haven’t been corrected for during gain normalization. I would try processing your gain reference using remove_outlier_pixels, which is a command line utility included with CisTEM (that has worked for me in the past).


1 Like