Table 2.

ROI biases and slopes

Model	CME Bias (HU)	NGE Bias (HU)
Aorta
UNet-Baseline	$-48[\pm 105](-\mathbf{1.39})$	$-15[\pm 29](-0.98)$
UNet-Phase	$-\mathbf{47}[\pm 74](-\mathbf{1.45})$	$1[\pm 27](-1.19)$
Pix2Pix	$14[\pm 119](-\mathbf{1.16})$	$\mathbf{30}[\pm 25](-0.57)$
CycleGAN	$-\mathbf{178}[\pm 67](-\mathbf{1.95})$	$-\mathbf{24}[\pm 38](0.05)$
Cortex
UNet-Baseline	$-\mathbf{50}[\pm 39](-\mathbf{1.86})$	$-\mathbf{39}[\pm 36](-\mathbf{1.59})$
UNet-Phase	$-15[\pm 39](-\mathbf{1.84})$	$-6[\pm 38](-\mathbf{1.50})$
Pix2Pix	$15[\pm 52](-\mathbf{1.51})$	$21[\pm 46](-1.08)$
CycleGAN	$-\mathbf{62}[\pm 51](-\mathbf{1.78})$	$-\mathbf{55}[\pm 45](-0.93)$
Medulla
UNet-Baseline	$-\mathbf{25}[\pm 23](-\mathbf{1.79})$	$-\mathbf{21}[\pm 26](-\mathbf{1.60})$
UNet-Phase	$7[\pm 18](-\mathbf{1.83})$	$9[\pm 19](-\mathbf{1.59})$
Pix2Pix	$\mathbf{31}[\pm 33](-\mathbf{1.48})$	$\mathbf{33}[\pm 31](-1.10)$
CycleGAN	$-22[\pm 56](-\mathbf{1.66})$	$-27[\pm 51](-0.62)$
Tumour
UNet-Baseline	$1[\pm 43](-1.40)$	$13[\pm 41](-1.32)$
UNet-Phase	$\mathbf{29}[\pm 44](-1.23)$	$\mathbf{37}[\pm 45](-0.87)$
Pix2Pix	$\mathbf{61}[\pm 45](-1.05)$	$\mathbf{66}[\pm 47](-0.55)$
CycleGAN	$5[\pm 61](-0.85)$	$3[\pm 51](-0.40)$

Data format: mean bias [95% LoA] (slope coefficient)

Bold indicates a significant proportional or constant bias for slope coefficient and mean bias, respectively