THREE-DIMENSIONAL TONGUE MOTION IN SPEECH

言语中的三维舌头运动

• 批准号: 6261236
• 负责人: MAUREEN STONE
• 金额: $ 45.74万
• 依托单位: UNIVERSITY OF MARYLAND BALTIMORE
• 依托单位国家: 美国
• 财政年份: 1993
• 资助国家: 美国
• 起止时间: 1993-07-01 至 2006-05-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6261236
• 关键词: bioimaging /biomedical imaging; biomechanics; clinical research; human subject; image processing; magnetic resonance imaging; model design /development; physical model; speech; tongue

This project will visualize internal and surface 3D- tongue motion during speech based on ultrasound and tagged Cine- MRI data. The project will use kinematic, dynamic and geometric models to capture, manipulate, and reduce the dimensionality of the major patterns underlying tongue behavior during normal speech production. These normative data can then be compared with patient populations. Three projects will be executed. The first will reconstruct the movement of 3D tongue surfaces and volumes during speech. We have already determined a minimal, optimal set of slices for reconstructing 3D surface motion from 2D ultrasound slices. The proposed experiments will add to this work by reconstructing 3D motion of internal tissue points from tagged-MRI. The observed internal motion will allow us, for the first time, to complement EMG by inferring tongue muscle activity from MRI. The value of such data is twofold. First, visualization of a complex speech movement patterns allow us to identify and quantify underlying patterns of movement and shape. Second, the data are useful clinically for comparing normal and disordered speech patterns, and estimating the underlying musculature. The second project, which is related to the first, will classify tongue motion based on biologically important tongue surface features. Geometric shape modeling will be done on 3D tongue surface shapes and motions. Geometric models extract global and local motion features that reduce the dimensionality of 3D structures, and identify and classify underlying patterns. This will be particularly useful for determining group and individual subject behavior patterns, and removing noise. In order to test hypotheses of muscle-to-surface linkages, the third project will develop a family of 3D mechanical tongue models. We have already developed 2D (one slice over time) kinematic and dynamic models of tongue motion. Our kinematic models represent surface and internal deformation of the tongue and reflect its underlying physiology. These models can fill gaps in the reconstructed 3D motions (Aim 1) and will provide input and insight into the dynamic predictive models. Dynamic finite element models will link internal muscular activity to surface shape motion. The 3D models will be physiologically accurate and can be made more or less complex to suit the experimental condition.

该项目将基于超声波和标记的 Cine-MRI 数据，将语音期间的内部和表面 3D 舌头运动可视化。 该项目将使用运动学、动态和几何模型来捕获、操纵和减少正常言语产生过程中舌头行为的主要模式的维度。 然后可以将这些规范数据与患者群体进行比较。 将实施三个项目。 第一个将重建言语过程中 3D 舌面和音量的运动。 我们已经确定了用于从 2D 超声切片重建 3D 表面运动的最小、最佳切片集。 拟议的实验将通过从标记 MRI 重建内部组织点的 3D 运动来补充这项工作。 观察到的内部运动将使我们第一次能够通过 MRI 推断舌肌活动来补充肌电图。 此类数据的价值是双重的。 首先，复杂语音运动模式的可视化使我们能够识别和量化运动和形状的潜在模式。其次，这些数据在临床上可用于比较正常和紊乱的言语模式，并估计潜在的肌肉组织。第二个项目与第一个项目相关，将根据生物学上重要的舌头表面特征对舌头运动进行分类。 几何形状建模将在 3D 舌头表面形状和运动上进行。 几何模型提取全局和局部运动特征，以降低 3D 结构的维度，并识别和分类底层模式。 这对于确定群体和个体受试者行为模式以及消除噪音特别有用。为了测试肌肉与表面联系的假设，第三个项目将开发一系列 3D 机械舌头模型。 我们已经开发了舌头运动的 2D（随时间变化的切片）运动学和动态模型。 我们的运动学模型代表了舌头的表面和内部变形，并反映了其潜在的生理学。 这些模型可以填补重建 3D 运动中的空白（目标 1），并将为动态预测模型提供输入和见解。 动态有限元模型将内部肌肉活动与表面形状运动联系起来。 3D 模型在生理上是准确的，并且可以变得或多或少复杂以适应实验条件。