Table 2.

Reasons for including each class in the ground-truth segmentations of the magnetic resonance images of the vocal tract during speech.

Class	Reason(s) for inclusion
Head	Primary: segmentation of the posterior pharyngeal wall would enable automatic measurement of the distance between the soft palate and the posterior pharyngeal wall
	Secondary: segmentation of the upper lip would enable automatic lip motion tracking
Soft palate	Segmentation would enable soft palate shape and motion analysis, and also automatic measurement of the distance between the soft palate and the posterior pharyngeal wall
Jaw	Segmentation of the lower lip would enable automatic lip motion tracking
Tongue	Segmentation would enable tongue shape and motion analysis
Vocal tract	Segmentation would enable vocal tract shape analysis
Tooth space	Included so that there were no holes in the ground-truth segmentations, thus facilitating the post-processing of estimated segmentations