Table 3.

Summary of heatmap similarity metrics for all configurations compared to the reference model (A2)

Input images	Id	Binarizer %	RMSE	Pearson correlation	MSSIM	EMD	IoU Top 40% R	IoU Top 10% R
Aligned T1w	A1	None	10.78 ± 3.76 [5.98, 21.47]	0.13 ± 0.04 [0.06, 0.21]	0.25 ± 0.08 [0.12, 0.42]	8.44 ± 3.25 [3.99, 17.48]	0.06 ± 0.02 [0.02, 0.11]	0.01 ± 0.01 [0.00, 0.03]
	B1	13.75	3.85 ± 0.87 [2.25, 5.48]	0.08 ± 0.03 [0.03, 0.15]	0.87 ± 0.04 [0.78, 0.94]	0.50 ± 0.29 [0.14, 1.29]	0.02 ± 0.01 [0.01, 0.04]	0.00 ± 0.00 [0.00, 0.01]
	C1	27.50	4.77 ± 0.96 [3.07, 6.64]	0.18 ± 0.05 [0.08, 0.26]	0.83 ± 0.04 [0.74, 0.89]	0.71 ± 0.45 [0.11, 1.71]	0.07 ± 0.02 [0.03, 0.10]	0.01 ± 0.01 [0.00, 0.02]
	D1	41.25	6.17 ± 1.15 [3.98, 8.34]	0.23 ± 0.05 [0.13, 0.30]	0.72 ± 0.06 [0.62, 0.83]	1.61 ± 0.62 [0.54, 2.75]	0.10 ± 0.02 [0.05, 0.14]	0.01 ± 0.01 [0.00, 0.03]
Skull-stripped T1w	A2	None	0.00 ± 0.00	1.00 ± 0.00	1.00 ± 0.00	0.00 ± 0.00	1.00 ± 0.00	1.00 ± 0.00
	B2	13.75	2.79 ± 0.49 [1.97, 3.79]	0.58 ± 0.07 [0.44, 0.69]	0.91 ± 0.03 [0.82, 0.96]	0.39 ± 0.25 [0.12, 1.05]	0.24 ± 0.06 [0.13, 0.35]	0.15 ± 0.03 [0.08, 0.21]
	C2	27.50	3.54 ± 0.61 [2.45, 4.85]	0.56 ± 0.07 [0.39, 0.67]	0.86 ± 0.04 [0.78, 0.92]	0.73 ± 0.44 [0.13, 1.67]	0.31 ± 0.05 [0.20, 0.37]	0.17 ± 0.03 [0.10, 0.24]
	D2	41.25	3.72 ± 0.59 [2.69, 4.96]	0.51 ± 0.08 [0.33, 0.65]	0.80 ± 0.05 [0.68, 0.89]	1.01 ± 0.50 [0.23, 2.12]	0.26 ± 0.03 [0.18, 0.31]	0.16 ± 0.04 [0.09, 0.22]

Values between brackets show the 95% confidence interval

RMSE root mean square error, MSSIM mean structural similarity index measure, EMD Earth Mover’s distance, IoU intersection over union, R relevance

Values in bold highlight best result for metric