The tongue serves vital oral and oropharyngeal sensorimotor functions and is crucial for such activities as swallowing and speech in everyday living. Tongue weakness and motor discoordination have been identified as the primary causes of dysphagia and dysarthria in a great number of diseases. Because of its unique anatomical complexity, current knowledge of the structure-function relationships of the in vivo human tongue remains poor. Our work aims at filling this knowledge void with accurate, reliable quantitative data that provide the foundation for evidence-based treatment planning and can serve as predictors for the effects of aging, disease, and rehabilitation. Using advanced imaging techniques, we undertake a multifaceted yet systematic approach toward achieving this major scientific objective, integrating anatomical, kinematic, kinetic, volumetric, and hemodynamic perspectives. Examples of the practical biomedical applications of this work include image-based lingual tumor staging, exercise-induced changes in lingual tissue composition and hemodynamics, and anatomical and biomechanical correlates of lingual muscle weakness versus motor discoordination in a variety of neurological diseases. Our major accomplishments in 2006 include the following:? ? A comprehensive study of task-induced changes in lingual blood flow and tissue kinematics has been completed using advanced 2D power Doppler, 3D power Doppler, and tissue velocity ultrasonography in three subject groups (healthy young, healthy senior, and neurologic patients). Our salient findings are: (1) aging is associated with a significant reduction in task-induced changes in lingual reperfusion peak systole, end diastole, acceleration, time averaged mean velocity, volume flow, and total reperfusion duration (p less than .0001 in all parameters); (2) swallows and effortful swallows, in contrast to maximum contraction tasks, are characterized by significantly less percent change during lingual reperfusion in nearly all hemodynamic parameters (p less than .0001); (3) duration of maximum oral contraction task significantly affects changes in 3D vascularization and flow indices during post-contraction reperfusion in the healthy young: longer task, greater change (p less than .0001); (4) magnitude of post-contraction peak systole in the tongue is highly linearly correlated with the magnitude of tongue pressure applied; (5) the tissue displacement and velocity profiles of the healthy seniors indicate the recruitment of floor-of-mouth muscles as a compensatory strategy to augment lingual tissue response to task demand.? ? """"""""Effortful swallow"""""""" is commonly used in swallowing rehabilitation. Clinical implementation of this compensatory strategy is inconsistent in method and results. With the goal of identifying the underlying mechanisms of task execution so as to improve clinical outcomes, we have completed a project on ?3D Lingual Strain Distribution at the Height of Effortful Swallow? using tagged MRI with simultaneous pneumographic monitoring of the magnitude of task-required maximum tongue effort. For data analysis, we have developed an innovative method for tagline detection based on pseudo wavelet decomposition with automatic indexing and linking procedures to correct artifacts and tagline discontinuity. In addition, we have used normalization weights, calculated from the corresponding pneumogram signals, to improve the accuracy of displacement maps. At present, we are assembling normalized displacement maps into 3D volumes and computing strain tensor for each voxel. Analysis is expected to be completed in 3 months with manuscripts to follow.? ? For lingual tissue characterization based on MRI T1 and T2 relaxation times, our in vivo pilot studies using a single-slice approach in 2005 showed significant regional differences in lingual tissue types. This year we devoted considerable effort in improving our imaging techniques bby exploring 3D-acquisition pulse sequences. We have succeeded in identifying the DESPOT1 and DESPOT2 imaging methods as better alternatives. These new methods permit us to achieve a 1-mm3 in vivo isotropic resolution, covering the entire tongue volume in less than 2 minutes. Data acquisition is being implemented. We anticipate significant increase in the accuracy of our measurements and T1 and T2 maps for detailed lingual tissue characterization. Further, we forecast the eventual dissemination of an imaging protocol suitable for inclusion in routine clinical evaluation of lingual tissue changes for quantitative monitoring of tumor progression and infiltration and tissue responses to radiotherapy or chemotherapy.? ? Using high resolution diffusion tensor imaging without gap in calf tongue models, we have completed a pioneering study and successfully mapped the complex 3D morphological characteristics of the entire volume of the tongue with accompanying gross anatomical dissection for cross validation. We developed a novel method to regularize the acquired image data based on normalized convolution with additional skewness similarity measures. Intrinsic and extrinsic lingual muscle compartments were successfully segmented according to tensor coherence defined as the product of angular, magnitude, and skewness similarity measures. In addition to characterizing DTI-based morphological details of lingual muscles, our imaging and analysis methods were sensitive enough to document (1) anterior vs. posterior differences in lingual tissue composition, and (2) interdigitation of fibers in selected muscle groups (e.g., radially oriented fibers in midst of longitudinally oriented fibers) in an interesting and distinctive tensor configuration with mixed prolate and oblate diffusion displacements. With technical advances that significantly reduce scan time, minimize susceptibility artifacts, and increase SNR and field homogeneity, DTI holds great promises for in vivo detection of morphological changes induced by disease, aging, or exercise.? ? For our volumetric evaluation of the tongue, we have collected a large quantity of multi-plane whole-tongue MR images during contraction tasks. Accurate volumetric analysis requires an optimal method of tongue segmentation. Having developed novel, prerequisite methods of high-speed image registration and distance-weighted multi-planar image fusion in 2005, this year we have focused on the development of a 2D semi-automatic segmentation program that deforms an operator-initialized rough polygon and converges to the actual tongue boundary. Currently, our algorithm conducts anisotropic smoothing, calculates image gradients, applies the SigMod function to generate a speed image, and uses the Geodesic Active Contour Level Set method to evolve the tongue boundary. This 2D semi-automatic segmentation program has proven to be considerably more efficient and reliable than the time-consuming and error-prone manual tracing. We forecast further improvement in tongue MRI segmentation performance and accuracy in the next phase of our development and image analysis efforts through modeling the tongue domain constraints, developing a trainable segmentation algorithm, fusing multi-plane volumetric MRI datasets, and extending the segmentation from 2D to 3D.? ? To optimize MR imaging, we have developed a 12-channel tongue MRI coil, the first of its kind, through extramural collaboration. We have overcome several mechanical issues, and our coil has been approved by the MRI Safety Subcommittee of the NIH MRI Facility for use with human subjects. Our tests thus far have shown superior SNR. At present we are evaluating and updating the coil in terms of imaging penetration and anticipate a final device that will enable parallel imaging and enhancement in image contrast and spatial resolution.
