Chewing cycle kinematics in young children using a low-cost 3D camera system: A proof-of-concept study

To analyze the feasibility of a low-cost three-dimensional (3D) tracking system to obtain spatial and temporal data, two-dimensional (2D) and 3D views and identify changes in jaw movements in young children in response to different portion sizes of food. Chewing cycle kinematics has been scarcely studied in young children largely due to the reliance on costly laboratory-based systems.

Jaw motion was recorded in five children (12-36 months of age) with an Intel™ RealSense D435 depth camera. Two paper markers were used (forehead and chin) while children chewed a small or large portion of Cheerios™. Two custom-codes were developed in Python™. The first allows manual selection of the markers and tracks motion during recording generating two output files for each sequence providing time-stamped 3D coordinates of the tracked markers. The second custom code enables offline views and post-processing of the data. Descriptive comparisons were performed for kinematic variables (e.g., cycle duration, maximum vertical and lateral displacement).

The adapted low-cost 3D camera system dependably recorded marker trajectories during chewing. Synchronized video and 3D coordinate data were obtained for each sequence, allowing offline views and selection of chewing cycles. The adapted system captured individual changes in chewing kinematics between the small and large portions, although the pattern of response was not uniform across variables/participants.

This proof-of-concept study supports the feasibility that the low-cost 3D camera system can capture chewing kinematic data in children aged 12- to 36-months and is sensitive enough to detect changes when chewing different portion sizes.