The results of previous work that dealt with group-wise registration and globally optimal models were clearly better than those obtained from pair-wise registration under similar conditions. This comes with a cost however. The improved process was far slower and the results were not superior by a very noticeable measure. In fact, it is quite possible that the results were partially biased due to the conditions set for the experiments.
Impending experiments aim to disprove the claim that the conditions and choices made were the cause for the apparent advantage. They just as well aspire to increase the apparent gain of group-wise registration as it is not yet evident. Various improvements and lines of research have already been suggested and similar work continues even at present.