Markerless Laryngeal Motion Tracking During Swallowing Using an RGB-D Camera with 3D Head-and-Neck Alignment

Tomoya Onishi, Yoshihiro Midoh, Jun Shiomi, Toru Yuba, Noriyuki Miura

Confirmation: I have read and agree with the IEEE BHI 2025 conference submission's policy on behalf of myself and my co-authors.

Keywords: aspiration, swallowing function test, depth sensor, markerless motion tracking, virtual 3D model

TL;DR: We propose a novel non-invasive method to monitor laryngeal motion using RGB-D imaging for early dysphagia detection in older adults.

Abstract: Dysphagia, or difficulty in swallowing, can lead to aspiration pneumonia—a lung infection caused by inhalation of food or fluids— particularly in older adults, making its early detection and prevention critically important. While invasive techniques such as X-ray fluoroscopy are traditionally used to evaluate swallowing function, there is a growing need for simple and noninvasive methods are needed not only for screening but also for monitoring the effectiveness of swallowing rehabilitation exercises such as effortful swallowing or vocal training. In this study, we propose a markerless laryngeal motion tracking method using a noncontact RGB-D camera combined with 3D head–neck model alignment. A subject-specific 3D model is constructed from depth images acquired in a static state, and swallowing-related laryngeal motion is extracted by compensating for forward–backward body sway and vertical neck movement through virtual model manipulation. In experiments using a water drinking test, the proposed method effectively reduced motion artifacts from body and neck movement and enabled clearer extraction of laryngeal motion associated with swallowing. A comparison across three subjects revealed consistent trends in skin displacement patterns corresponding to muscle activity during swallowing.

Track: 7. General Track

Registration Id: SXNGRQS6FS5

Submission Number: 184