Using Sequence Analyses to Quantitatively Measure Oropharyngeal Swallowing Temporality in Point-of-Care Ultrasound Examinations: A Pilot Study

Abstract

(1) Background: Swallowing is a complex process that comprises well-timed control of oropharyngeal and laryngeal structures to achieve airway protection and swallowing efficiency. To understand its temporality, previous research adopted adherence measures and revealed obligatory pairs in healthy swallows and the effect of aging and bolus type on the variability of event timing and order. This study aimed to (i) propose a systemic conceptualization of swallowing physiology, (ii) apply sequence analyses, a set of information-theoretic and bioinformatic methods, to quantify and characterize swallowing temporality, and (iii) investigate the effect of aging and dysphagia on the quantified variables using sequence analyses measures. (2) Method: Forty-three participants (17 young adults, 15 older adults, and 11 dysphagic adults) underwent B-mode ultrasound swallowing examinations at the mid-sagittal plane of the submental region. The onset, maximum, and offset states of hyoid bone displacement, geniohyoid muscle contraction, and tongue base retraction were identified and sorted to form sequences which were analyzed using an inventory of sequence analytic techniques; namely, overlap coefficients, Shannon entropy, and longest common subsequence algorithms. (3) Results: The concurrency of movement sequence was found to be significantly impacted by aging and dysphagia. Swallowing sequence variability was also found to be reduced with age and the presence of dysphagia (H(2) = 52.253, p < 0.001, η² = 0.260). Four obligatory sequences were identified, and high adherence was also indicated in two previously reported pairs. These results provided preliminary support for the validity of sequence analyses for quantifying swallowing sequence temporality. (4) Conclusions: A systemic conceptualization of human deglutition permits a multi-level quantitative analysis of swallowing physiology. Sequence analyses are a set of promising quantitative measurement techniques for point-of-care ultrasound (POCUS) swallowing examinations and outcome measures for swallowing rehabilitation and evaluation of associated physiological conditions, such as sarcopenia. Findings in the current study revealed physiological differences among healthy young, healthy older, and dysphagic adults. They also helped lay the groundwork for future AI-assisted dysphagia assessment and outcome measures using POCUSs. Arguably, the proposed conceptualization and analyses are also modality-independent measures that can potentially be generalized for other instrumental swallowing assessment modalities.

Keywords: aging; deglutition; dysphagia; information theory; ultrasonography.

Figure A1

Event Order Probability Matrices of Swallowing Event Pairs across Groups. Event pair order: first event on left, second event on top; matrices: top—healthy young adults (number of permuted sequences m_p = 8309), middle—healthy older adults (number of permuted sequences m_p = 3281), bottom—dysphagic adults (number of permuted sequences m_p = 249).

Figure 1

A systemic conceptualization of swallowing physiology at multiple levels.

Figure 2

(a) Placement of the gel pad. (b) Placement of the transducer. (c) Annotated frame sample of USI acquired at the submental region on the mid-sagittal plane. Structures of interest: the surface of tongue base shown as a hyperechoic contour, hyoid bone as a reflective, acoustic shadow, and the geniohyoid muscle as a hypoechoic polygon.

Figure 3

The Swallow Trial Protocol.

Figure 4

(a) A visual presentation of permuting identical timestamps. (b) Schematic representation of data extraction and analysis workflow.

Figure 5

Annotated and extracted video frame sample from a healthy young participant (top), the corresponding concurrent event timelines in the form of a Gantt chart (middle), and the permuted possible event sequences (bottom).

Figure 6

Mean overlap coefficients of different event transition pairs (error bars showing 95% Confidence Interval) across groups for overlapping (a) hyoid bone displacement and geniohyoid contraction initiation (HB-GH_init), (b) sustained hyoid bone displacement and geniohyoid contraction (HB-GH_sus), (c) tongue base retraction and geniohyoid contraction initiation (TB-GH_init), (d) sustained tongue base retraction and geniohyoid contraction (TB-GH_sus), (e) tongue base retraction and hyoid bone displacement initiation (TB-HB_init), and (f) sustained tongue base retraction and hyoid bone displacement (TB-HB_sus).

Figure 7

Video frame extracts annotated with event labels (top), possible sequences after permutation (middle), obligatory sequences found in the healthy young population (bottom).