Cost functions can use either direct intensity information or landmark information. For each type of cost function there are two main categories: intra-modal (for example, sum of absolute differences) and inter-modal (for example, mutual information). Not surprisingly, it is significantly harder to find functions with desirable properties for inter-modal registration and consequently much research has been concerned with finding such functions. For this report it is the intensity-based inter-modal cost functions that are investigated: for example, Mutual Information, Correlation Ratio, etc. These are more difficult to optimise, since there are many non-linear, potentially discontinuous terms involved that result in functions that are less smooth and regular. In general, though, all cost functions (except some landmark-based cost functions which are convex) require global optimisation.