Incorporating Dental Anatomy Into MRI-Based Vocal Tract Models Using Zero Echo Time Imaging

Accurate modeling of the vocal tract requires faithful representation of both soft tissue and bony structures, particularly the teeth and mandible, which directly shape the oral airway and influence speech acoustics. Conventional MRI-based vocal tract models typically exclude teeth due to the inability of standard MRI sequences to image short-T2 tissues, while computed tomography (CT), though effective for bone, involves ionizing radiation and is unsuitable for many speech research applications.

To evaluate zero echo time (ZTE) MRI as a radiation-free method for imaging dental and mandibular structures relevant to vocal tract modeling, and to assess the geometric and acoustic consequences of incorporating ZTE-derived teeth into MRI-based vocal tract models.

Twenty-five participants underwent upper-airway ultra-low-dose CT and MRI, including ZTE-MRI and conventional gradient-echo (GRE) acquisitions. Dental structures were segmented from ZTE-MRI and CT and fused with GRE-derived vocal tract airways to construct hybrid ZTE+GRE-MRI and hybrid CT+GRE-MRI models. For three professional voice users, vocal tract area functions were computed for GRE-only and hybrid models and analyzed using a transmission-line model-based acoustic simulation to estimate vocal tract transfer functions for sustained vowel /a/.

ZTE-MRI provided clear visualization of teeth and mandibular structures with pseudo-CT-like contrast. In cases with dental hardware (eg, fillings, implants, and bridges), ZTE-MRI showed localized signal nulling, while CT exhibited more extensive streaking artifacts. Hybrid ZTE+GRE models closely matched CT-based models (area differences ∼1.7-2.3%), whereas GRE-only models overestimated anterior oral cavity airspace (up to ∼16-17%). These geometric differences translated to measurable shifts in acoustic output, particularly in higher-order formant frequencies.

ZTE-MRI enables anatomically complete vocal tract modeling, including supporting radiation-free visualization of dental structures. ZTE-derived teeth produce geometric and acoustic effects comparable to those derived from CT, addressing a key limitation of conventional MRI-based speech modeling and offering a practical alternative for speech and voice research, particularly in populations where radiation exposure is undesirable.