Tracking tongue motion in three dimensions using tagged MR image

Harmonic phase (HARP) analysis has been used in tagged magnetic resonance imaging (MRI) to measure two-dimensional (2D) in-plane motion and strain, and was recently applied in characterizing the motion of tongue during speech. The 3D-HARP method extended the HARP method to track three-dimensional (3D) cardiac motion from short- and long-axis tagged MR images by diffusing 2D in-plane motion on a sparse 3D mesh. In this paper, we propose a new 3D-HARP method to calculate the 3D tongue motion from three orthogonal tag orientations imaged within sagittal, coronal, and axial image planes. Our method iteratively tracks 2D in-plane motions on points on a compact mesh using 2D-HARP followed by thin-plate spline (TPS) interpolation to extend the 2D motion to the whole mesh. Experiments on real tongue data show that our method is capable of accurate motion tracking in simple utterances

谐波相（HARP）分析已用于标记的磁共振成像（MRI），以测量二维（2D）的面内运动和应变，最近应用于表征语音期间舌头运动的表征。 3D - 毛线方法扩展了HARP方法，以通过在稀疏的3D网格上扩散2D平面运动，从短轴和长轴标记的MR图像跟踪三维（3D）心脏运动。在本文中，我们提出了一种新的3D - 毛线方法，以从矢状，冠状和轴向图像平面内成像的三个正交标签方向计算3D舌头运动。我们的方法迭代地跟踪紧凑型网格上点上的2D内部运动，然后使用2D毛线进行薄板样条（TPS）插值，以将2D运动扩展到整个网格。真实舌头数据的实验表明，我们的方法能够以简单的话语进行准确的运动跟踪