Using Active Shape Modeling Based on MRI to Study Morphologic and Pitch-Related Functional Changes Affecting Vocal Structures and the Airway

Research output: Contribution to journalArticle

6 Citations (Scopus)
31 Downloads (Pure)

Abstract

Objective: The shape of the vocal tract and associated structures (eg, tongue and velum) is complicated and varies according to development and function. This variability challenges interpretation of voice experiments. Quantifying differences between shapes and understanding how vocal structures move in relation to each other is difficult using traditional linear and angle measurements. With statistical shape models, shape can be characterized in terms of independent modes of variation. Here, we build an active shape model (ASM) to assess morphologic and pitch-related functional changes affecting vocal structures and the airway. Method: Using a cross-sectional study design, we obtained six midsagittal magnetic resonance images from 10 healthy adults (five men and five women) at rest, while breathing out, and while listening to, and humming low and high notes. Eighty landmark points were chosen to define the shape of interest and an ASM was built using these (60) images. Principal component analysis was used to identify independent modes of variation, and statistical analysis was performed using one-way repeated-measures analysis of variance. Results: Twenty modes of variation were identified with modes 1 and 2 accounting for half the total variance. Modes 1 and 9 were significantly associated with humming low and high notes (P < 0.001) and showed coordinated changes affecting the cervical spine, vocal structures, and airway. Mode 2 highlighted wide structural variations between subjects. Conclusion: This study highlights the potential of active shape modeling to advance understanding of factors underlying morphologic and pitch-related functional variations affecting vocal structures and the airway in health and disease.
Original languageEnglish
Pages (from-to)554-564
Number of pages11
JournalJournal of Voice
Volume28
Issue number5
Early online date12 Mar 2014
DOIs
Publication statusPublished - Sep 2014

Fingerprint

Statistical Models
Principal Component Analysis
Tongue
Analysis of Variance
Respiration
Spine
Magnetic Resonance Spectroscopy
Cross-Sectional Studies
Health

Keywords

  • MRI
  • shape
  • voice
  • pitch
  • hum
  • statistical modelling
  • image analysis
  • active shape model
  • active appearance model
  • vocal tract
  • humming
  • cervical spine
  • posture

Cite this

@article{57d7263ba4b74e5b86927b3fe4f28146,
title = "Using Active Shape Modeling Based on MRI to Study Morphologic and Pitch-Related Functional Changes Affecting Vocal Structures and the Airway",
abstract = "Objective: The shape of the vocal tract and associated structures (eg, tongue and velum) is complicated and varies according to development and function. This variability challenges interpretation of voice experiments. Quantifying differences between shapes and understanding how vocal structures move in relation to each other is difficult using traditional linear and angle measurements. With statistical shape models, shape can be characterized in terms of independent modes of variation. Here, we build an active shape model (ASM) to assess morphologic and pitch-related functional changes affecting vocal structures and the airway. Method: Using a cross-sectional study design, we obtained six midsagittal magnetic resonance images from 10 healthy adults (five men and five women) at rest, while breathing out, and while listening to, and humming low and high notes. Eighty landmark points were chosen to define the shape of interest and an ASM was built using these (60) images. Principal component analysis was used to identify independent modes of variation, and statistical analysis was performed using one-way repeated-measures analysis of variance. Results: Twenty modes of variation were identified with modes 1 and 2 accounting for half the total variance. Modes 1 and 9 were significantly associated with humming low and high notes (P < 0.001) and showed coordinated changes affecting the cervical spine, vocal structures, and airway. Mode 2 highlighted wide structural variations between subjects. Conclusion: This study highlights the potential of active shape modeling to advance understanding of factors underlying morphologic and pitch-related functional variations affecting vocal structures and the airway in health and disease.",
keywords = "MRI, shape, voice, pitch, hum, statistical modelling, image analysis, active shape model, active appearance model, vocal tract, humming, cervical spine, posture",
author = "Nicola Miller and Jenny Gregory and Aspden, {Richard Malcolm} and Pete Stollery and Gilbert, {Fiona Jane}",
note = "Copyright {\circledC} 2013 The Voice Foundation. Published by Mosby, Inc. All rights reserved.",
year = "2014",
month = "9",
doi = "10.1016/j.jvoice.2013.12.002",
language = "English",
volume = "28",
pages = "554--564",
journal = "Journal of Voice",
issn = "0892-1997",
publisher = "Mosby Inc.",
number = "5",

}

TY - JOUR

T1 - Using Active Shape Modeling Based on MRI to Study Morphologic and Pitch-Related Functional Changes Affecting Vocal Structures and the Airway

AU - Miller, Nicola

AU - Gregory, Jenny

AU - Aspden, Richard Malcolm

AU - Stollery, Pete

AU - Gilbert, Fiona Jane

N1 - Copyright © 2013 The Voice Foundation. Published by Mosby, Inc. All rights reserved.

PY - 2014/9

Y1 - 2014/9

N2 - Objective: The shape of the vocal tract and associated structures (eg, tongue and velum) is complicated and varies according to development and function. This variability challenges interpretation of voice experiments. Quantifying differences between shapes and understanding how vocal structures move in relation to each other is difficult using traditional linear and angle measurements. With statistical shape models, shape can be characterized in terms of independent modes of variation. Here, we build an active shape model (ASM) to assess morphologic and pitch-related functional changes affecting vocal structures and the airway. Method: Using a cross-sectional study design, we obtained six midsagittal magnetic resonance images from 10 healthy adults (five men and five women) at rest, while breathing out, and while listening to, and humming low and high notes. Eighty landmark points were chosen to define the shape of interest and an ASM was built using these (60) images. Principal component analysis was used to identify independent modes of variation, and statistical analysis was performed using one-way repeated-measures analysis of variance. Results: Twenty modes of variation were identified with modes 1 and 2 accounting for half the total variance. Modes 1 and 9 were significantly associated with humming low and high notes (P < 0.001) and showed coordinated changes affecting the cervical spine, vocal structures, and airway. Mode 2 highlighted wide structural variations between subjects. Conclusion: This study highlights the potential of active shape modeling to advance understanding of factors underlying morphologic and pitch-related functional variations affecting vocal structures and the airway in health and disease.

AB - Objective: The shape of the vocal tract and associated structures (eg, tongue and velum) is complicated and varies according to development and function. This variability challenges interpretation of voice experiments. Quantifying differences between shapes and understanding how vocal structures move in relation to each other is difficult using traditional linear and angle measurements. With statistical shape models, shape can be characterized in terms of independent modes of variation. Here, we build an active shape model (ASM) to assess morphologic and pitch-related functional changes affecting vocal structures and the airway. Method: Using a cross-sectional study design, we obtained six midsagittal magnetic resonance images from 10 healthy adults (five men and five women) at rest, while breathing out, and while listening to, and humming low and high notes. Eighty landmark points were chosen to define the shape of interest and an ASM was built using these (60) images. Principal component analysis was used to identify independent modes of variation, and statistical analysis was performed using one-way repeated-measures analysis of variance. Results: Twenty modes of variation were identified with modes 1 and 2 accounting for half the total variance. Modes 1 and 9 were significantly associated with humming low and high notes (P < 0.001) and showed coordinated changes affecting the cervical spine, vocal structures, and airway. Mode 2 highlighted wide structural variations between subjects. Conclusion: This study highlights the potential of active shape modeling to advance understanding of factors underlying morphologic and pitch-related functional variations affecting vocal structures and the airway in health and disease.

KW - MRI

KW - shape

KW - voice

KW - pitch

KW - hum

KW - statistical modelling

KW - image analysis

KW - active shape model

KW - active appearance model

KW - vocal tract

KW - humming

KW - cervical spine

KW - posture

U2 - 10.1016/j.jvoice.2013.12.002

DO - 10.1016/j.jvoice.2013.12.002

M3 - Article

VL - 28

SP - 554

EP - 564

JO - Journal of Voice

JF - Journal of Voice

SN - 0892-1997

IS - 5

ER -