Table 3.

Transformation outcomes in independent complementary testing sets (Step 2).

	Training sets (Step 1)		Independent testing sets (Step 2)
	Size (N)	Iterations	Size (N)	ICC mean	ICC SD	ICC 2.5Q–97.5Q		ICC ≥ 0.8 (%)
FTLD-Tau GM	24	100	15	0.95	0.02	0.90	0.98	100
	12	100	27	0.95	0.01	0.92	0.97	100
FTLD-Tau WM	24	100	15	0.95	0.02	0.91	0.98	100
	12	100	27	0.95	0.01	0.92	0.96	100
FTLD-TDP GM	24	100	29	0.82	0.05	0.69	0.91	72
	12	100	41	0.81	0.06	0.69	0.89	70
FTLD-TDP WM	24	100	29	0.86	0.03	0.80	0.91	98
	12	100	41	0.85	0.03	0.78	0.89	95

FTLD-Tau, frontotemporal lobar degeneration with inclusions of the tau protein; FTLD-TDP, frontotemporal lobar degeneration with inclusions of the transactive response DNA-binding protein 43 kDa; ICC, intraclass correlation coefficient; GM, gray matter; N, number of tissue samples; Q, quantile; SD, standard deviation; WM, white matter. We performed 100 iterations of linear regression in training sets of N = 12 and N = 24 sample size, and applied equivalence factors to independent testing sets including all remaining tissue samples in each iteration. Here, we report transformation outcomes (i.e., ICC) in these complementary testing sets. We report mean, standard deviation and a non-parametric quantile-based confidence interval (2.5–97.5% of the distribution) for the ICC. Additionally, we report the frequency of optimal transformations out of 100 iterations per group based on our threshold of ICC ≥ 0.8.