Table 4.

TRE between known and recovered alignment of the simulated slices, averaged over simulations from 3 stacks, 6 stacks and all simulations. Last column presents p-value for all simulations, showing statistical significance of performance compared to the full method.

	3 Stacks (mm)	6 Stacks (mm)	All (mm)	p-Value
Full	0.83	0.72	0.77
Huber	0.92	0.74	0.83	0.039
No robust	1.05	0.80	0.95	0.001
No matching	0.96	0.78	0.87	0.015
Reference	0.76	0.65	0.72	0.160