Next Experiments

1. Look at results for intensity OF's.
2. For better intensity OF's, choose all possible shape OF's and see results.
3. Maybe also do the reverse: choose shape OF's that work well and try the full range of intensity OF's. (1), (2) and (3) have been implicitly done already. Some of the better methods for shape/intensity have been used while 'scanning' for another complementary objective function.
4. Try to see how weight affects the different registration methods. Does it have optimal values that differ depending on the objective function devised?
5. Try different optimisers. There are still optimisation duration issues which might be addressed.
6. Continue trying a multi-resolution approach and an increase in knot-points.
7. Change knot-points number at different stages of NRR instead of keeping it constant.
8. Try different data for all the experiments above.
9. Look at different slices together, for example by moving up and down while showing the discrepancy image.
10. Try some group-wise regimes and see effects.