Sensitivity
To compare the different quality measures in more detail, we define sensitivity as the mean slope of a measure over the range investigated, divided by the mean uncertainty in the measure - that is the smallest misregistration that can just be detected. The sensitivities of the different methods evaluated are plotted below, demonstrating that the model-based approach with a shuffle radius between 1.5 and 2.4 is the most sensitive.
5#5 = 31#31