
62 Statistical Computing in Nuclear Imaging
2.4.2 OTHER METHODS
T
he Bayes risk discussed in the previous section uses the loss function. In
this section we briefly introduce two other approaches that can be used in
BE condition fo r s ystem o ptimization (experimental de sign) that are based
on computatio n of some utility function that quantifies the difference between
the prior a nd the posterior. Since the loss function is not utilized it is not
optimal from the decision-theoretic p erspective. However, these two methods
are quite straightforward to implement and are useful for situations where the
loss function is unknown or cannot be reliably specified. ...