Introduction
Swallowing, which is the final stage of oral processing, ensures smooth transport of the orally processed food towards the stomach for further digestion [
1]. Human swallowing is an involuntary action that takes place about 1000 times a day [
2]. Pharyngeal swallowing is important to study since the pharynx is partly shared by the airways and food swallowing tract [
3]. Misdirection of the bolus at this stage means that the bolus enters the airways, resulting in aspiration and possibly, pneumonia. Swallowing disorders (dysphagia) are a growing concern and it is estimated that dysphagia affects roughly 8% of the world population [
4]. The physiological responses of people who are suffering from dysphagia caused by neurological conditions or age-related impairment are insufficient to handle the rapid flow of foods or liquids through the oropharynx. Therefore, thickeners, which are typically gum- or starch-based, are added to the food to slow down the flow of the bolus. Starch-based thickener swells upon hydration while gum-based thickeners form network thereby holding water and increasing the viscosity. A thickener, whether gum- or starch-based shear thins during flow, i.e., the perception of thickness decreases with increasing speed of deformation [
5]. This necessitates accurate information of shear rate during swallowing. European Society for Swallowing Disorders (ESSD) in its recently published White Paper, has stressed the importance of rheological parameters such as shear rate, non-Newtonian fluids properties, yield stress, elasticity, and density [
6]. To test these parameters in humans is not only cumbersome but it also possesses ethical issues. For example in our earlier work, aimed at relating elasticity and safe swallowing, it was noticed that even though high elasticity in liquids has an effect on safe swallowing, the results were not statistically significant unless a large number of patients were evaluated and variability among subjects was kept small [
7].
The assessment tool used by clinicians is either manometry or video-fluoroscopy [
8,
9]. Manometry is clinically vital but technically hard to perform [
10]. While executing manometry, distinction between hydrodynamic pressure, bolus pressure as it touches transducer and contact pressure, measured pressure as pharyngeal wall touches the transducer must be made [
10]. Furthermore, manometry is based on the insertion of a probe into the patient’s pharynx, which obstructs the bolus flow [
8,
9] and causes discomfort. During video-fluoroscopic analysis, the swallowing of fluids is monitored using X-ray imaging and the entire swallowing process is recorded, therefore enabling the examiner to follow the swallowing sequence frame by frame [
11]. However, video-fluoroscopy necessitates the use of radio-opaque contrast media that has been to shown to alter the rheology of the bolus [
12].
To analyze different parameters suggested in the White Paper by ESSD, we propose bolus flow measurement using ultrasound velocity profiling (UVP) technique, which detects the movement of suspended particles/bubbles in a flowing liquid using Doppler echography. The technique, which is non-invasive and can measure velocity profiles in real time, can be applied to measure the velocity profile of the bolus and thereby determine accurate shear rate during swallowing. The UVP technique is described in detail elsewhere [
13].
The swallowing process has been simulated in vitro by Mackley et al. [
14] and Noh et al. [
15] previously (existing models are reviewed in [
16]). The model presented by Mackley et al. named “The Cambridge throat” does not simulate the epiglottis movement during bolus flow hence mimicking only severe dysphagia. The Cambridge throat did not report the shear rate and manometry information that is vital to study the influence of rheology with respect to swallowing. Nevertheless, the Cambridge throat presents a good starting point for in vitro simulation of swallowing process. The in vitro model presented by Noh et al. used video-fluoroscopy to follow the bolus flow thus mimicking in vivo equivalent of swallowing analysis. The model published by Noh et al. lacks epiglottis structure and does not report shear rate and manometry information that dysphagia community is most interested in, as mentioned earlier. Both these models conclude that a thickened bolus travels slower in the oropharynx and too thick consistency bolus leaves post-swallow residues as also noticed in clinical studies [
4]. Moreover, several theoretical and simulation studies have been performed on the swallowing process [
17‐
21]. Simulation studies normally assume highly idealized conditions, such as uniform bolus geometry and Newtonian fluids. Considering that most boluses, especially the ones thickened for dysphagia are shear thinning and often elastic, means that at least non-Newtonian fluids have to be considered. Furthermore, the complex geometry and peristaltic type of motion exposing the bolus to complex sequence of shear and extension deformation makes mathematical modeling even more challenging [
16]. For detailed description of existing in vitro swallowing models, the readers are advised to the book chapter [
22]. In our opinion, an in vitro simulator that performs in vivo type of analysis without interfering with the human body provides the best compromise between the two extremes of modeling and clinical studies.
Therefore, the aim of the present work was to design an in vitro device that mimics human swallowing, and that could be used to study the rheological parameters suggested in the White Paper by ESSD. Furthermore, the device should allow studies of the flow properties of the bolus and to simulate both healthy swallowing and various swallowing disorders. In this work, we presented for the first time the shear rate distribution during bolus swallowing using non-invasive UVP technique. Additionally, the physiologically realistic geometry was utilized to perform in vitro manometry.
Materials and Methods
Materials
Commercial rapeseed oil was used as a model Newtonian fluid (ICA-Maxi, Sweden) and was compared to the shear-thinning fluid Fresubin Clear (Fresenius-Kabi GmbH, Bad Homburg, Germany), which is a thickener used to manage dysphagia. Rapeseed oil was used as is, whereas Fresubin Clear was mixed to syrup consistency by adding 3.91 g of the powder to 100 ml of water.
Methods
The shear viscosity of the samples was determined with the ARES-G2 rheometer (TA Instruments, New Castle, DE, USA), using a cone-plate geometry with plate diameter of 40 mm and cone angle of 0.04 rad. Sample density was measured with the Densito 30PX density meter (Mettler Toledo AB, Stockholm, Sweden).
Flow Visualization and Verification Using Ultrasonics
Flow visualization was performed using the UVP technique. In this work, a commercial UVP system that included electronics, software, and transducers was used (Incipientus™ Ultrasound Flow Technologies AB, Gothenburg, Sweden).
Experimental Loop for UVP and Manometry in a Continuous Flow
To validate that measurements can be performed in the given geometry with ultrasound and pressure transducers, initially the fluids were continuously pumped from a tank using a rotatory lobe pump (Sterilobe SLAS; Johnsons Pumps, Lanarkshire, UK) operated at two different speeds, corresponding to two different flow rates (Tables
2,
3).
Table 2
Percent difference between flow rates measured with the gravimetric method and velocity profile integration (UVP) methods, shear rate at 50 s−1, and the physical properties of the test fluids
Rapeseed oil | 0.97 | 0.89 | 8.14 | 1.15 | 0.063 |
2.36 | 2.21 | 6.1 | 6.4 | |
Fresubin Clear | 1.14 | 0.98 | 13.65 | 43.78 | 0.65 |
2.78 | 2.06 | 25.8 | 38.45 | |
Table 3
In vitro shear rate and corresponding maximum velocities recorded
1 | 0.22 | 124 |
2 | 0.161 | 78.83 |
3 | 0.162 | 67.6 |
4 | 0.144 | 71 |
Average | 0.170 | 85.36 |
Standard deviation | 0.031 | 39.76 |
The entire model was filled with the fluid, and the valves on the trachea and nasal cavity were closed. The ultrasound transducer was attached using a specially designed holder to ensure firm contact between the transducer face and the model. For optimal transmission of the acoustic waves, rapeseed oil was introduced between the transducer and the model pharynx block as coupling media. Several commercially available acoustic media were tested before and rapeseed oil was finally selected due to better matching of acoustic impedance with the given material used in the model pharynx. The temperature of the fluid was strictly monitored during the experiments within 20.5–22.0 °C.
Validation of the manometry was performed during continuous flow using the pressure drop between the lower, mid-pharynx pressure transducer (see Fig.
1), and the outlet (atmospheric pressure), and comparing this to the calculated pressure drop using the Hagen–Poiseuille law and the gravimetrically measured mass flow rate. In order to eliminate geometrical effects, the pressure drop is presented in a simplified form as the ratio of the pressure at a high-flow rate (
PH) to the pressure at a low-flow rate (
PL):
\(\frac{{(P_{\text{H}} - P_{0} )}}{{(P_{\text{L}} - P_{0} )}}.\)
Pump speeds corresponding to the flow rates reported in Table
2, where
P0 is the atmospheric pressure at the fluid exit point, were applied.
Similarly, the flow rates are expressed as the ratio of the high-flow rate (
\(\dot{Q}_{\text{H}}\)) to the low-flow rate (
\(\dot{Q}_{\text{L}}\)):
\(\left( {\frac{{\dot{Q}_{\text{H}} }}{{\dot{Q}_{\text{L}} }}} \right)^{n}\) with
n representing the power law coefficient (
n = 1 for the Newtonian oil, and
n < 1 for the shear-thinning thickener solution). According to the Hagen–Poiseuille law, the changes in pressure drop and flow rates should be equal as in Eq.
1:
$$\frac{{(P_{\text{H}} - P_{0} )}}{{(P_{\text{L}} - P_{0} )}} = \left( {\frac{{\dot{Q}_{\text{H}} }}{{\dot{Q}_{\text{L}} }}} \right)^{n} .$$
(1)
The data were acquired when the pressure values reached a steady state, as monitored by the PicoScope software (Pico Technology, Cambridgeshire, England). The measurements were repeated three times and the standard deviation from the mean value did not exceed 2.5%.
Methodology for UVP Measurements
The ultrasonic beam passes through the polycarbonate material of the model into the actual fluid flow (Fig.
1) which means that it will be refracted at each interface. The Doppler angle, i.e., the angle between the fluid flow and ultrasonic beam, inside the model cavity was measured using a reference 90° ultrasound beam. The two values were utilized to determine the angle inside the model pharynx for each test fluid.
An average of 128 velocity profiles was used to determine the volume flow rate and shear rate. A Doppler angle of 60° and ellipse short-axis radius 8.4 mm was determined, while the sound velocities in rapeseed and Fresubin Clear were 1443.6 m/s and 1560 m/s, respectively. The base frequency of the non-invasive ultrasound transducer was 5 MHz, and 5 cycles/pulse were used for velocity profile measurements.
Shear Rate Calculation from the Velocity Profile
Shear rate distribution (Eq.
2) from the wall to the center
\(\dot{\gamma }\) is calculated from the gradient of velocity profiles (
v), recorded with the UVP device as
$$\dot{\gamma } = - \frac{{{\text{d}}v(r)}}{{{\text{d}}r}}.$$
(2)
To capture the velocity gradient, a second-order polynomial was fitted to the velocity profiles recorded with the UVP. The highest shear rate is the one calculated from the velocity gradient at the wall of the model pharynx.
This method of shear rate measurement is well established and documented in many published articles as [
13,
30‐
32]. The shear rate presented is a measure on four bolus injections in the model cavity. Each data set representing an average of 128 velocity profiles is processed individually with UVP software.
Reference Measurement
The total volumetric flow rate through a cylinder with elliptical cross-section for a Newtonian fluid is given by Eq.
3 as found in literature [
33]
$$Q = \frac{\pi }{2}v a b,$$
(3)
where
v is the maximum velocity at the center of the tube, and
a and
b are the axes of the ellipse, in this case 8.4 mm and 18.2 mm, respectively.
To compute the volumetric flow rate (\(\dot{Q}\)) through a cross-section of the pharynx, the cross-section is first broken up into n segments, each with area An. The velocity in each segment is denoted Vn.
Then, assuming negligible secondary flow, the volumetric flow rate can be calculated using Eq.
4:
$$Q = \sum\limits_{n = 1}^{N} {\left( {V_{n} A_{n} } \right)} .$$
(4)
The conventional Bucket and StopWatch (BSW) method was used as a reference measurement. Thus, the mass flow rate was measured as a function of time by filling a bucket with the test fluid while recording time using a stopwatch. The mass flow rate was then converted to the volumetric flow rate (
\(\dot{V}\)) using the relation
\(\dot{V} = \frac{{\dot{m}}}{\rho },\) where
m is the mass and
ρ is the density of the test fluid. These parameters were measured with a high-precision electronic scale (PG4002-S; Mettler Toledo AB, Stockholm, Sweden) and a digital density meter. This procedure was repeated at least five times for each sample with two flow rates of the pump. The average value was used as the reference and compared with the flow rate determined using UVP (Eq.
4) and the error difference percentage was calculated.
In Vitro Bolus Injection Procedure
Final bolus injection settings were chosen based on initial pre-experiment. The main criterion was to ensure a smooth flow of a bolus of defined volume (15 ml in this case) released by the syringe. The UES was closed after 3 s after the bolus departed from the pharynx. The final volume of liquid considered for analysis is the one collected after the UES closing the esophagus pipe and that volume varied 1.2 ml around the mean value of 15 ml.