Publication Type | Academic Article |
Authors | Kim S, Barnett A, Pierpaoli C, Chi-Fishman G |
Journal | NMR Biomed |
Volume | 21 |
Issue | 5 |
Pagination | 479-88 |
Date Published | 06/01/2008 |
ISSN | 0952-3480 |
Keywords | Diffusion Magnetic Resonance Imaging, Imaging, Three-Dimensional, Muscle Fibers, Skeletal, Signal Processing, Computer-Assisted, Tongue |
Abstract | This study was performed to assess the feasibility of investigating the complex lingual myoarchitecture through segmentation of muscles from diffusion tensor imaging (DTI) data. The primary eigenvectors were found to be adequate for delineating the superior and inferior longitudinalis, genioglossus, and hyoglossus. The tertiary eigenvector orientations effectively revealed the homogeneous and systematic change of muscle orientation in the tongue core. In the longitudinalis near the tongue tip, the secondary eigenvectors were oriented in the radial direction. Lingual muscles were segmented using two methods: modified directional correlation (DC) and tensor coherence (TC) methods. The DC method, based on one eigenvector, was found to be inadequate for lingual muscle segmentation, whereas the TC method, based on the tensor shape and orientation, was used successfully to segment most lingual muscles. The segmentation result was used to report the diffusion tensor properties of individual lingual muscles. Also found was a continuous change in skewness of the intrinsic tongue core from negative in the anterior region to positive in the posterior region. DTI and the proposed segmentation method provide an adequate means of imaging and visualizing the complex, compartmentalized musculature of the tongue. The potential for in vivo research and clinical applications is demonstrated. |
DOI | 10.1002/nbm.1215 |
PubMed ID | 17952877 |