Should not be a problem… Extract as usual, you need to have a large box size to include high resolution information. You can use tight inner-mask size to focus on the center of particle (maybe 0.7-0.8 as a start). 2D classification should be straightforward, and as you “thin out the herd” and end up with small subsets of particles in certain conformations or of highest resolution content, the particles used in the reconstruction will be separated well from others, and the near-neighbors will cancel as background noise. Highest-resolution structures of apoferritin all have each molecule immediately touching the neighbor. The major consideration is ice thickness, you want to be certain to have a single sheet of tightly packed particles, and not particles that are overlapping in the Z-plane (from top of scope to bottom, they should not be shallower and deeper and “under” each other). For this you should attempt thinner ice with more force or time in the blotting or amenable buffer/detergent, or be certain to pick only the middle of the hole, which has thinnest ice.