Table 3.

Estimated sample sizes (95% CI) in comparison to the ideal case of no-atrophy using whole brain annualised PBVC (80% power at the 5% significance level to detect 25% reduction in disease progression). Below, it is shown the percentage difference of sample size with the 95% CI and paired t-test between methods.

N = 41	FSL	LEAP	NLMI	Prop. non-Longitudinal	Prop. Longitudinal
Sample size	80 (52 to 125)	80 (50 to 129)	85 (54 to 134)	77 (51 to 117)	75 (46 to 125)

Percentage difference in mean (95% CI), p-value
FSL vs LEAP	-0.2 (-22.7 to 22.2), p = 0.98
FSL vs NLMI	6.3 (-14.0 to 26.8), p = 0.53
FSL vs Prop. non-Longitudinal	-4.0 (-15.9 to 7.9), p = 0.51
LEAP vs NLMI	6.6 (-12.0 to 25.3), p = 0.48
LEAP vs Prop. non-Longitudinal	-3.8 (-22.7 to 15.1), p = 0.69
NLMI vs Prop. non-Longitudinal	-9.8 (-26.0 to 6.4), p = 0.23
FSL vs Prop. Longitudinal	-5.6 (-38.4 to 27.0), p = 0.74
LEAP vs Prop. Longitudinal	-5.4 (-31.8 to 21.9), p = 0.69
NMLI vs Prop. Longitudinal	-11.3 (-34.8 to 12.2), p = 0.34
P. non-Longitudinal vs Prop. Longitudinal	-5.9 (-33.6 to 21.7), p = 0.67