High-resolution dynamic speech imaging with deformation estimation.

Area of Science:

• Medical Imaging
• Speech Science
• Biomedical Engineering

Background:

• Dynamic speech magnetic resonance imaging (DSMRI) offers real-time visualization of articulatory motion.
• Existing DSMRI methods face limitations in imaging speed and quantitative motion analysis.
• Accurate articulatory motion data is crucial for speech production research and clinical applications.

Purpose of the Study:

• To develop a novel DS-MRI technique for simultaneous estimation of dynamic speech image sequences and deformation fields.
• To enhance spatial resolution and frame rate for improved visualization of articulatory dynamics.
• To enable direct quantitative analysis of articulatory motion using derived deformation fields.

Main Methods:

• Extension of previous Partial Separability (PS) model-based methods for DS-MRI.
• Simultaneous estimation of dynamic image sequences and associated deformation fields.
• Achieved spatial resolution of 2.2 × 2.2 mm² and a nominal frame rate of 100 fps.

Main Results:

• The proposed technique successfully visualizes speech motion and deformation with high spatial resolution and frame rate.
• Deformation fields derived from the technique allow for direct quantitative analysis of articulatory motion.
• In vivo experiments demonstrated the effectiveness and practical utility of the method.

Conclusions:

• The novel DS-MRI technique significantly advances real-time articulatory motion visualization and analysis.
• High-resolution dynamic imaging and deformation field estimation enable detailed phonetic studies, as shown with Brazilian Portuguese.
• This method holds potential for improved understanding of speech production and clinical diagnostics.