SNR Equation
where the constant K includes hardware dependent factors such as coil, pre-amp and noise power spectrum, field strength dependent factors, and pulse sequence type parameters (TR, TE etc), and tissue dependent parameters (spin density, T1, T2). This expands to
for 2D SNR where K is a constant, FOVx and FOVy are the field-of-view in the x and y directions, Nx and Ny are the number of frequency and phase encoding steps, Δz is the slice thickness, NEX is the number of signal averages, and BW is the receive bandwidth.
If your SNR gets to be above 20:1, any further increase is not likely to produce a benefit to the observer. In this case, you'd be better off making savings in the scan time or increasing the matrix size instead.
Further reading on this topic:
Books: MRI From Picture to Proton p70-71, Q&A in MRI p97-99, MRI The Basics p167-169
Online: Case Physics