Neutrinos are tiny, have no charge and are extremely weakly interactive. Detecting neutrinos relies upon the extremely unlikely event of a neutrino striking the nucleus of an atom directly resulting in the emission of light.
In crystals atoms are arranged in patterns. When we look through models of crystals we notice that at certain angles whole rows of atoms may be hidden from view because they are obscured by the first atom in the row. If a neutrino was to pass through the crystal at this angle it would be extremely unlikely that it would strike a nucleus.
However, we can also imagine that if we alter the angle that we are looking at the crystal by the very smallest amount then the atoms behind the first in the row gradually come into view. Thus, it must be true, that if a neutrino where to pass through a crystal there are certain angles of incidence that would result in a greater chance that it strike a nucleus.
If we could grow crystals of an appropriate material (ice?) large and ordered enough, then perhaps we could arrange them in such a way so as to maximize the potential for a collision along one particular angle of incidence at the expense of the rest and thus build a neutrino detector that could be pointed at potential sources of the particle.