Related Experiment Video
Updated: May 24, 2025
10:13
Hemi-laryngeal Setup for Studying Vocal Fold Vibration in Three Dimensions
Published on: November 25, 2017
10.9K
GIRAFE: Glottal imaging dataset for advanced segmentation, analysis, and facilitative playbacks evaluation.
Gustavo Andrade-Miranda1,2, Konstantinos Chatzipapas1,3, Julián D Arias-Londoño4
1Laboratoire de Traitement de l'Information Médicale (LaTIM), UMR 1101, INSERM, University of Brest, 29200, Brest, France.
Data in Brief
|March 3, 2025
Summary
A new dataset, GIRAFE, addresses the lack of annotated vocal fold images for developing advanced glottal gap segmentation techniques. This resource aids research in high-speed videoendoscopic analysis and facilitative playback evaluation.
Area of Science:
- Medical Imaging
- Bioengineering
- Vocal Fold Dynamics
Background:
- Facilitative playback development from high-speed videoendoscopic sequences is limited by scarce, annotated vocal fold datasets.
- Lack of semantic segmentation data for the glottal gap hinders reproducibility and research exploration in this field.
Purpose of the Study:
- To introduce GIRAFE, a novel data repository for vocal fold analysis.
- To facilitate the development of advanced semantic segmentation and analysis techniques for high-speed videoendoscopic sequences.
Main Methods:
- Compilation of 65 high-speed videoendoscopic recordings from 50 patients (healthy, voice disorder, unknown condition).
- Manual annotation of glottal gap semantic segmentation masks by an expert.
- Inclusion of automatic glottal area segmentation using state-of-the-art methods.
Main Results:
- The GIRAFE dataset supports studies on glottal gap segmentation algorithms.
- It aids in improving and creating new facilitative playback techniques from high-speed videoendoscopic data.
Conclusions:
- GIRAFE addresses the critical need for annotated vocal fold imaging data.
- The challenge of fully automatic, accurate glottal area semantic segmentation remains an open research area.
