I may be wrong, but i dont think that Q=|B/G|....the general definition of Q is that it equals w*( energy stored/avg power dissipated)
Consider the example of a parallel RLC circuit. Here Y = G + j/wL (w^2*L*C - 1) and Q = w0 * (1/2*C*((I*R)^2))/(1/2 * I^2 * R) = w0*R*C.
Hopefully it clarifies or else let me know if something is wrong
Y=G+jB refers to a parallel equivalent circuit. Consider an ideal capacitor or inductor with a parallel loss conductance. Obviously a high quality factor requires the conductance G to be low related to the susceptance B. So Q=|B/G| is right. But it only applies to simple lossy L or C elements, not to RLC circuits.