Using atlas-driven imaging for determining variations in velopharyngeal function among children with cleft palate and hypernasal speech Grant uri icon

abstract

  • The broad overarching goal of this study is to develop a highly innovative tool to identify the physiologic causes of abnormal velopharyngeal function, as seen in children with hypernasal speech. First, using a normative noncleft population at 5 to 8 years of age, we will establish a magnetic resonance imaging (MRI) method and a spatiotemporal atlas to visualize 3D muscle movements during representative speech samples. Our approach will include consideration of internal variables that have been shown to impact velopharyngeal function, including race, sex, and dialect [8, 10-12]. The main drawback of speech MRI has been the slow imaging speed and serial sampling of data points in the traditional one-image-at-a-time dynamic time series. Recent literature in dynamic real-time MRI of speech has used a wide range of protocols, with frame rates ranging from 2 frames (or images) per second (FPS) up to ~20 FPS [13]. These rates are far from providing adequate temporal resolution during speech, where a complete transitional movement may take as little as 100 ms, requiring at least 30-40 FPS to resolve the motion [14]. Recently, we have developed ultrafast dynamic imaging techniques based on a Partial Separability (PS) framework, enabling us to sample the entire vocal tract in 3D at 166 FPS [15, 16]. Two critical challenges remain: shortening the acquisition to make it suitable for young children and automating the extraction of clinically relevant information from motions in the large number of imaging volumes generated over a speech sample. This proposal aims to overcome these challenges and create a tool for speech imaging to define the physiologic causes of hypernasality. Secondly, we will apply the atlas to two groups of children with repaired cleft palate--those with hypernasal speech (secondary to velopharyngeal dysfunction) and those with normal resonance--to determine the physiologic variances that separate the two groups and that separate the groups from healthy controls. Similar to our observation of differences in the anatomy [9] among those with repaired cleft palate, we hypothesize deviations from normal will be evident in both patient groups. However, we hypothesize that those with hypernasal speech will present with different patterns of insufficient motion, poor initial geometry, or changes in timing coordination of articulatory events.
  • This is a supplement award tied to a parent R01 award.

date/time interval

  • December 2020 - June 2025