5th International Conference on Spoken Language Processing

Sydney, Australia
November 30 - December 4, 1998

Segmentation Of The Airway From The Surrounding Tissues On Magnetic Resonance Images: A Comparative Study

Alain Soquet (1), Véronique Lecuit (1), Thierry Metens (2), Bruno Nazarian (1), Didier Demolin (3)

(1) Laboratoire de Phonetique Experimentale, Université Libre de Bruxelles, Belgium
(2) Unite O.R.L. Hopital Erasme, Belgium
(3) Laboratoire de Phonologie Experimentale, Université Libre de Bruxelles, Belgium

Magnetic Resonance Imaging techniques are uniquely attractive in their ability to provide an extensive body of information on the vocal tract geometry. Once the images are acquired, they must be further processed in order to segment the airway from the surrounding tissues, so as to locate the air passage. This problem has been addressed in several ways i n the literature. In this paper, we carry out a comparative study of different approaches to the same body of data in order to assess the accuracy of the different methods. It is shown that the different methods present small average error and large error distribution.

Full Paper

Bibliographic reference.  Soquet, Alain / Lecuit, Véronique / Metens, Thierry / Nazarian, Bruno / Demolin, Didier (1998): "Segmentation of the airway from the surrounding tissues on magnetic resonance images: a comparative study", In ICSLP-1998, paper 0531.