Designing drug occupancy studies with PET neuroimaging: Sample size, occupancy ranges and analytical methods

•Likelihood-based approaches yield considerably better VT-based IC50 estimates than those derived from the Lassen plot.•We see very limited benefit in scanning individuals given drug doses that correspond to less than 40% target occupancy.•Within the range of typical PET sample sizes (520) each additional subject clearly reduced the error in the IC50 estimate.•We introduce a new method for direct estimation of IC50 which was less sensitive to noisy plasma drug concentrations.