Related Experiment Videos
Three-dimensional tongue surface reconstruction: practical considerations for ultrasound data
1Department of Computer Science, Johns Hopkins University, Baltimore, Maryland 21218, USA. lundberg@cs.jhu.edu
The Journal of the Acoustical Society of America
|November 26, 1999
Summary
This study presents a method for reconstructing 3D tongue surfaces using fewer ultrasound slices, achieving high accuracy for speech sound analysis. This approach simplifies data collection for 3D tongue imaging.
Area of Science:
- Speech Science
- Biomedical Engineering
- Medical Imaging
Background:
- Three-dimensional (3D) tongue surface reconstruction is crucial for understanding speech production.
- Previous methods utilized 60 ultrasound slices, which can be impractical for dynamic speech analysis.
Observation:
- This study aimed to identify a minimal, sparse set of ultrasound slices for accurate 3D tongue reconstruction.
- The method was optimized using coarse-to-fine search on coronal slices and validated using midsagittal contours.
Findings:
- Sparse reconstructions achieved maximum errors under 1.5 mm and standard deviation errors under 0.32 mm, with 80% surface coverage.
- Optimizing slice selection from midsagittal contours yielded comparable results to using a full coronal set, demonstrating generalizability across subjects.
Implications:
- This optimized sparse data acquisition method reduces the complexity and impracticality of collecting ultrasound data for 3D tongue imaging.
- The findings support the development of more efficient and accessible tools for speech research and clinical applications.