Table 2.

Dice coefficients ($\mu \pm \sigma$) for validation (10 datasets) and test (20 datasets) sets. Left: Each network in the framework was trained and evaluated using the expert ground truth. Right: the cascade framework was executed, and in the intermediate stages the results from the previous stages were employed. Only the final results were compared to the expert ground truth

Mask	$dsc\_\mu \pm dsc\_\sigma$
	Independent		Cascade
	Validation set	Test set	Validation set	Test set
Pharynx	$0.89\pm 0.01$	$0.87\pm 0.02$	$0.89\pm 0.03$	$0.87\pm 0.02$
Tongue	$0.91\pm 0.03$	$0.89\pm 0.04$	$0.91\pm 0.02$	$0.89\pm 0.03$
Soft palate	$0.84\pm 0.06$	$0.79\pm 0.09$	$0.839\pm 0.05$	$0.79\pm 0.08$