Abstract
Objective: The goal of this paper is to assess the validity of various metrics developed to characterize tongue shapes and positions collected through ultrasound imaging in experimental setups where the probe is not constrained relative to the subject’s head. Patients and Methods: Midsagittal contours were generated using an articulatory-acoustic model of the vocal tract. Sections of the tongue were extracted to simulate ultrasound imaging. Various transformations were applied to the tongue contours in order to simulate ultrasound probe displacements: vertical displacement, horizontal displacement, and rotation. The proposed data analysis method reshapes tongue contours into triangles and then extracts measures of angles, x and y coordinates of the highest point of the tongue, curvature degree, and curvature position. Results: Parameters related to the absolute tongue position (tongue height and front/back position) are more sensitive to horizontal and vertical displacements of the probe, whereas parameters related to tongue curvature are less sensitive to such displacements. Conclusion: Because of their robustness to probe displacements, parameters related to tongue shape (especially curvature) are particularly well suited to cases where the transducer is not constrained relative to the head (studies with clinical populations or children).