Automatic motion analysis of the tongue surface from ultrasound image sequences

Abstract

The authors present a system for automatic 2D analysis of the tongue movement from digital ultrasound image sequences. The system focuses on extraction, tracking and analysis of the tongue surface during speech production and swallowing. The input to the system is provided by a Head and Transducer Support System (HATS), which is developed for use in ultrasound imaging of tongue movement. The authors developed a novel active contour (snakes) model that uses several adjacent images during the extraction of the tongue surface contour for an image frame. The user supplies an initial contour model for a single image frame in the sequence. This initial contour is a form of expert knowledge input to the system, which is used to find the candidate contour points in the adjacent images. Subsequently, the new snake mechanism is applied to estimate optimal contours for each image frame using these candidate points. Finally, the system uses a postprocessing method to refine the positions of the contours by utilizing more spatiotemporal information. The authors extended their previous work by applying the system to different speech and swallowing sequences using various constraints. The extended system can also extract qualitative local deformations with only a minimal computational overhead which may be useful for the diagnosis of Cerebellar Ataxic disorder. The authors tested the system on several different speech and swallowing sequences produced by HATS. During the tests, the authors saw that the system is flexible enough to be used in a wide variety of cases. In addition, visual inspection of the detected contours by the speech experts confirms that the results are very promising and this system can be effectively employed in speech and swallowing research.