This protocol uses advanced magnetic resonance imaging and ultrasonography to address several important issues related to structure-function relationships of the human tongue. These include: (1) the compressibility of the human tongue and its common, yet untested, reference as a muscular hydrostat, (2) task-induced interactions between lingual musculature and vasculature and region-specific vascular demands; and (3) changes in lingual tissue characteristics, deformation and strain distribution, and fiber orientation as a function of contraction tasks in health, aging, disease, and exercise. By quantitatively address these issues, this protocol will elucidate the functional biomechanical intricacies of the in vivo human tongue and ultimately contribute to the development of innovative, biologically realistic, and physiologically sound treatment techniques methodologies for lingual sensorimotor impairments and help our patients regain functional movement and control of the tongue in swallowing and speech. This report period was marked by multiple achievements in the development of innovative methodologies and techniques for optimization of imaging and data analysis. Highlights of our major accomplishments and findings are as follows: a) Dependence of Apparent Diffusion Tensor on Diffusion Time in Muscle Tissue of Tongue and Heart. We have successfully completed our Diffusion Time project in which we used fresh ex vivo tongue and heart models to determine the time dependence of the apparent diffusion tensor (anisotropic diffusivity scalar) of muscle tissue in these organs. By measuring diffusion times - time delays between diffusion sensitizing gradient pulses - over a range (32-810 ms) achievable on a clinical scanner, we have discovered that such time dependence distinctly exists for lingual and cardiac muscles, in contrast to brain tissue where prohibitively short diffusion times must be used (< 6 ms). Dependence of diffusivity on diffusion time indicates the presence of barriers that restrict the otherwise free diffusion. This is the first research to substantiate diffusion time dependency for water molecules in muscular tissue. In addition, by fitting the eigenvalues of the apparent diffusion tensor of lingual and cardiac muscles to a novel two-compartment model, we have found that muscle fiber diameter size and volume fraction can be adequately estimated. These findings substantiate the usefulness of diffusion tensor imaging in our investigative direction that, as we believe, will eventually lead to a better understanding of the structure-function relationships of the human tongue. When coupled with technical advances that significantly reduce scan time, minimize susceptibility artifacts, and increase field homogeneity, DTI holds great promises for non-invasive, in vivo detection of changes in muscle myoarchitecture due to disease, treatment, aging, and exercise. b) Development and Validation of an Automatic Tagline Detection Method. We have completed the development and validation of an innovative, robust, automatic tagline detection method for the analysis of tissue motion from tagged MR images of the tongue. The motivation for this project was to achieve maximal precision in lingual tagline tracing, as conventional tagline detection methods fall short in accuracy for lingual strain analysis due to the large curvature of the tongue, its highly complex tissue architecture, and interference from surrounding structures. Our method is based on multi-resolution pseudo wavelets synthesis, coupled with automatic tagline clustering and fitting algorithms. Validation of this method and comparison with existing state-of-the-art methodology showed significant improvement in lingual tagline pixel displacement measurement (p = .0011). c) Development of 3-D Segmentation Methodology for MRI-Based Lingual Volumetric Analysis. We have completed Phases I and II of this development effort that aims at greatly improving the accuracy of tongue segmentation in both manual and computer-aided semi-automatic operations. In Phase I, we successfully developed the algorithm for image registration using shape-constrained mutual information. The novelty of our registration method lies in the integration of tongue shape into the mutual information metric and the development of a real-coded genetic algorithm that performs rapid optimization. Our validation experiments have demonstrated high speed, robustness and accuracy in image registration. In Phase II, we successful implemented and validated a distance-weighted multi-planar image fusion method for 3D volumetric image reconstruction using sagittal and coronal tongue MR scans. Phase III is in progression, and we have completed the feasibility testing of the level-set and active contour segmentation algorithms and comparative applications to 2D tongue MR images and synthetic dataset. Our persistent efforts are steadily advancing us toward a tongue-specific analysis method based on multi-resolution Gaussian pyramid and progressive level-set segmentation. d) Toward MRI-Based Quantitative Lingual Tissue Characterization: Studies on T1-T2 Relaxation Times. Founded on the fact that T1 and T2 relaxation times are sensitive to tissue structure and changes in biochemical composition, this pioneering project focuses on high resolution measurements of T1 and T2 relaxation times of lingual tissue and its constituents (muscular, adipose and connective tissue) at 3.0T and 1.5T both in vivo and ex vivo. Analyses to date show: (i) considerable similarities in tissue composition between the bovine and human tongues, and (ii) significant gender-based differences in humans. These findings support continued usage of the bovine model for refinement of imaging and analysis techniques, and more extensive quest into male-female differences in lingual fat deposition and metabolic changes. T1 and T2 measurements, to be coupled with diffusion-weighted measurement of water mobility and MR spectroscopy of metabolite concentrations in tissue, hold great promise for quantitative monitoring of tumor progression and infiltration in muscles and surrounding tissues, and tissue responses to radiotherapy or chemotherapy. e) Imaging Optimization. Our efforts to reduce susceptibility artifacts and improve field homogeneity for MR imaging of the tongue, which directly affect image contrast and spatial resolution, have resulted in: ? the design of an MRI coil that uses multiple parallel, electrically decoupled elements, thus boosting signal-to-noise ratio without limiting the optimal imaging plane; ? the pioneering development of an oral insert (?artificial palate?, similar in concept to the intraoral coil described in the protocol) using dental impression material, to fill the air space between the palate and the superior surface of the tongue, reducing MR field inhomogeneity by 65-85% during imaging of the tongue; ? the completion of feasibility testing for the use of custom pyrolytic graphite blocks (both continuously nucleated and substrate nucleated) as passive shims for tongue MR imaging, discovering that submental placement of the passive shims in optimized orientation doubles lingual tissue contrast and improves tissue delineation by twofold.
