Figure 8.

And the same tradeoff for the longitudinal in-vivo experiment. Clearly the population standard deviation has little effect, which is to be expected given that we are fitting a linear mixed effects model with a random intercept per subject. So the only variance we really care about is the within-subject variance. This simulation again assumes 10 subjects per group and four scans per subject. The black lines indicate standard deviations of absolute (solid) and relative (dashed) volumes as determined by our imaging data.