Skip to main content
NCBI home page
The Journal of Physiology logoLink to The Journal of Physiology
. 2004 Oct 7;561(Pt 2):597–610. doi: 10.1113/jphysiol.2004.073502

Pharyngeal airway wall mechanics using tagged magnetic resonance imaging during medial hypoglossal nerve stimulation in rats

Michael J Brennick 1, Stephen Pickup 2, Lawrence Dougherty 2, Jacqueline R Cater 3, Samuel T Kuna 4
PMCID: PMC1665366  PMID: 15579543

Abstract

To better understand pharyngeal airway mechanics as it relates to the pathogenesis and treatment of obstructive sleep apnoea, we have developed a novel application of magnetic resonance imaging (MRI) with non-invasive tissue tagging to measure pharyngeal wall tissue motion during active dilatation of the airway. Eleven anaesthetized Sprague-Dawley rats were surgically prepared with platinum electrodes for bilateral stimulation of the medial branch of the hypoglossus nerve that supplies motor output to the protrudor and intrinsic tongue muscles. Images of the pharyngeal airway were acquired before and during stimulation using a gated multislice, spoiled gradient recalled (SPGR) imaging protocol in a 4.7 T magnet. The tag pulses, applied before stimulation, created a grid pattern of magnetically imbedded dark lines that revealed tissue motion in images acquired during stimulation. Stimulation significantly increased cross-sectional area, and anteroposterior and lateral dimensions in the oropharyngeal and velopharyngeal airways when results were averaged across the rostral, mid- and caudal pharynx (P < 0.001). Customized software for tissue motion-tracking and finite element-analysis showed that changes in airway size were associated with ventral displacement of tissues in the ventral pharyngeal wall in the rostral, mid- and caudal pharyngeal regions (P < 0.0032) and ventral displacement of the lateral walls in the mid- and caudal regions (P < 0.0001). In addition, principal maximum stretch was significantly increased in the lateral walls (P < 0.023) in a ventral–lateral direction in the mid- and caudal pharyngeal regions and principal maximum compression (perpendicular to stretch) was significantly increased in the ventral walls in all regions (P < 0.0001). Stimulation did not cause lateral displacement of the lateral pharyngeal walls at any level. The results reveal that the increase in pharyngeal airway size resulting from stimulation of the medial branch of the hypoglossal nerve is predominantly due to ventral displacement of the ventral and lateral pharyngeal walls.

Obstructive sleep apnoea is a respiratory disorder in humans characterized by the repetitive closure of the pharyngeal airway during sleep. Previous imaging and physiological studies of the pharyngeal airway in animals and humans using fibre optics, computerized tomography (CT) and magnetic resonance imaging (MRI) have revealed important insights about the pathogenesis of obstructive sleep apnoea: (1) the pharyngeal airway in obstructive sleep apnoea patients is smaller and more collapsible than that in normal subjects (Horner et al. 1989b; Shepard et al. 1991; Isono et al. 1997; Schwartz et al. 1998; Schwab et al. 2003); and (2) activation of pharyngeal muscles dilates and stiffens the airway (Brouillette & Thach, 1979; Strohl & Fouke, 1985; Redline & Strohl, 1987; Strohl et al. 1987; Eisele et al. 1995; Hida et al. 1995; Fregosi & Fuller, 1997; Fuller et al. 1999). In an effort to determine the factors that control airway patency, investigators have measured changes in airway size and shape to indirectly assess the mechanical properties of the pharyngeal wall, i.e. surrounding pharyngeal soft tissues (Wheatley et al. 1991; Schwab et al. 1995; Ryan & Love, 1996; Brennick et al. 1998; Morrell et al. 1998). While studies using conventional CT and MRI studies can determine the volume of pharyngeal wall soft tissue structures, these imaging techniques are unable to directly evaluate the mechanical properties of these soft tissues. Our relatively limited understanding of pharyngeal mechanics potentially prevents the development of new effective treatments for obstructive sleep apnoea. For example, selective neuromuscular stimulation of pharyngeal muscles has been studied as a possible treatment to prevent airway closure by increasing pharyngeal muscle dilating forces during sleep (Schwartz et al. 1996; Eisele et al. 1997; Ilomaki et al. 1997; Oliven et al. 2001). Although previous studies in animals and humans have examined the changes in airway size and stiffness due to muscle stimulation (Fuller et al. 1999; Brennick et al. 2001; Kuna, 2001; Kuna & Brennick, 2002; Kuna, 2004), there is little information as to how activation of the pharyngeal muscles alters the soft tissues in the pharyngeal walls to effect these airway changes.

To address these limitations, we have adopted MRI in combination with spatial modulation of magnetization (SPAMM, developed at the University of Pennsylvania) to track tissue motion in the pharyngeal wall during muscle stimulation (Axel & Dougherty, 1989). The SPAMM technique uses a series of radio frequency (RF) and magnetic field gradient pulses to generate an evenly spaced grid of dark lines in the target tissues. These lines track with the tissues as they move during muscle stimulation. Acquisition of images before and during muscle contraction using a spoiled gradient recalled imaging (SPGR) MRI protocol results in a series of images in which the grid pattern is distorted due to the tissue motion. Analysis of the images using the distorted grid lines as fiducial markers yields detailed information about tissue motion in the pharyngeal walls. MRI with SPAMM tagging has been used extensively to study cardiac mechanics (Axel et al. 1992; Marcus et al. 1997; Gotte et al. 1999; Scott et al. 1999; Yuan et al. 2000) and there have been several reports using MRI and SPAMM to investigate tongue movements in relation to speech (Niitsu et al. 1994; Napadow et al. 1999; Stone et al. 2001). However, this is the first application of MRI with SPAMM to examine pharyngeal mechanics during selective muscle stimulation. Specifically, we examined how stimulation of the medial branch of the hypoglossus nerve, supplying motor output to the genioglossus, geniohyoid and intrinsic tongue muscles, affects the displacement and strain of tissues surrounding the pharyngeal airway and how this pharyngeal wall tissue motion relates to changes in airway size and shape.

Methods

Surgical preparation

All studies were conducted with the approval of the University of Pennsylvania Institutional Animal Care and Use Committee. The surgical and MRI protocols were carried out in 11 male Sprague-Dawley rats, mean weight 386 ± 9.5 g (s.e.m. given throughout), under urethane anaesthesia (1.8 g kg−1, i.p., supplemented with 0.2 g kg−1, i.p., as needed). Following the experiments, the anaesthetized rats were killed by intracardiac injection of KCl-saturated aqueous solution. Arterial oxygen saturation, heart rate and temperature were continuously monitored (VetOx 4700, Sensor Devices, Inc. Waukesha, WI, USA). Core temperature was maintained at 35–36°C using a heated water pad (T-Pump, Gaymar Ind., Orchard Park). An adequate depth of anaesthesia was determined by the absence of any response to strong pressure on the paw. With the rat supine, a ventral mid-line incision from the genu of the mandible to the sternum allowed for the placement of a tracheal cannula and exposure of the medial hypoglossal nerve bilaterally. The rat breathed spontaneously from a 100% oxygen flow-by connected to the tracheal cannula. The hypoglossal nerve was separated from surrounding tissues and cut at a point approximately 1.0–1.5 cm proximal to the bifurcation of its medial and lateral branches. The lateral branch was cut away from the nerve trunk, leaving the medial branch intact. The distal cut end of the nerve trunk was installed into a cuffed bipolar electrode that utilized 0.127 mm diameter Teflon-coated platinum wires with 1.0 mm exposure on the tips. These cuffed electrodes were used to stimulate the medial hypoglossal nerve (Brennick et al. 2001).

The bipolar platinum leads (15 cm long) from each cuff electrode were connected to 26 G PVC-insulated and shielded copper wires that extended to a constant current photo-isolated stimulus unit (Grass PSIU 8, Astro-Medical, Warwick) and stimulator (Grass S44, Astro-Medical). Two stimulators were connected in tandem to provide a single stimulation train pattern with independent adjustment of current levels for each bipolar electrode. All pulsed nerve stimulations were delivered at 90 Hz, 0.1 ms pulse duration. This frequency achieves fused contraction in upper airway muscles (Brennick et al. 2001). Following electrode placement, we determined initial current settings for each electrode as the minimum current that visually produced maximum tongue protrusion. Final settings were determined using an iterative process during imaging, as described below.

Imaging protocol

MRI was performed in a 40 cm bore, 4.7 T magnet (Magnex, Concord) with a Varian INOVA Console (Varian, Palo Alto) using a 12 cm shielded gradient insert capable of generating gradients up to 25 g cm−1. A 65 mm inside diameter Litz coil (Doty, Columbia) was used for transmission and reception. The supine rat rested on a Plexiglas platen that inserted into the RF probe so that the junction of the hard and soft palate of the rat (centre of the region of interest) was at the mid-point of the probe's 52 mm RF window, at the isocentre of the gradient coil. A sagittal scout image was used to position the rat and probe in the magnet.

The gated SPGR with SPAMM protocol proceeded as follows: (1) a trigger signal from the stimulator initiated the cycle followed by a delay (250–350 ms); (2) SPAMM pre-conditioning pulses (9.8 ms) were followed by the first multislice SPGR image acquisition (87 ms); (3) bilateral stimulation of the medial hypoglossal nerve was initiated and no images were acquired during the following 100 ms; and (4) the second multislice image series was acquired (87 ms) during nerve stimulation. Each SPGR axial image was a square matrix, 128 pixels × 128 pixels, field of view (FOV) 35 mm, slice thickness 1 mm, interslice gap 1 mm, repetition time (TR) 87 ms, echo time (TE) 2.2 ms, with four averages so that 512 cycles were required for image data acquisition.

We used an iterative method to determine the stimulus intensity level for right and left electrodes. Previous studies (Brennick et al. 2001) revealed that medial hypoglossal stimulation dilated both the oropharyngeal and velopharyngeal airways. Therefore, a preliminary series of axial images (with two instead of four averages) were acquired using the initial stimulus intensity settings (determined on the bench level from direct observation, as described above) and these images were carefully inspected on the console for two criteria: (1) stimulation causing oropharyngeal and velopharyngeal dilatation throughout the airway; and (2) left to right symmetry in axial slices during stimulation. The stimulator constant-current output to each electrode was adjusted to achieve these two criteria usually in one or two trials. In order to determine whether the desired stimulation intensity was obtained, another SPGR iteration was performed at twice the operating level. The lower stimulation level was considered optimal if stimulation with the greater current caused no further appreciable airway dilatation. If additional significant dilatation was observed, then the current was again doubled and compared to the previous level until no further increases in airway size were observed. It should be noted that, in order to preserve the electrode–nerve interface, the upward iterations of current were not tested when relatively small increases in airway size were observed with the higher current. The mean current for a single electrode (when averaged over all electrodes) for all rats was 0.29 ± 0.09 mA.

We monitored the stimulator outputs (trigger signal and stimulator currents) with a digital oscilloscope (Agilent, Model 5461B, Palo Alto). The stimulus train duration averaged 184 ± 7 ms. The time for a single pass of the gated MRI protocol was 430 ± 20 ms for the first three experiments and was increased to 610 ± 20 ms for the later eight experiments. After it was discovered in the first three experiments that the SPAMM lines would persist for a longer period, the imaging cycle time was increased to (1) reduce the possibility of muscle fatigue by reducing muscle duty cycle (stimulus on/stimulus off) from approximately 45% to 25.6%, and (2) improve the overall image quality by increasing repetition time (TR; see discussion on MRI protocol considerations).

Following the determination of stimulus intensity levels, a complete set of axial images was acquired in two offset (11 slice) series with 1.0 mm interslice gap that were then merged to form a contiguous (22 slice) axial series. Post-acquisition interleaving of the two axial series improved the signal to noise ratio by limiting signal loss that would occur if the series were initially acquired as one contiguous acquisition. The image data were interpolated to a 256 pixel × 256 pixel matrix and transferred to a Silicon Graphics Octane workstation (SGI, Mountain View) for analysis using the SPAMM visualization software package (SPAMMVU, University of Pennsylvania). Sagittal images, using the same MRI protocol with FOV of 80 mm on a 128 by 128 square matrix, were also obtained to determine tissue displacement orthogonal to the axial plane images.

Image analysis

For each rat, a series of 22 contiguous axial slices were obtained. These data sets were aligned using a method previously described (Brennick et al. 2001). Briefly, a single axial slice in the forebrain just caudal to the olfactory bulb was identified as having a brain cross-section that was significantly smaller than the previous caudal slice. This unique and identifiable alignment slice was approximately 3 mm rostral to the junction of the hard to soft palate. Once this slice was identified in the axial series of each rat, a one-to-one correspondence among axial slices among all rats was easily obtained and verified by matching anatomical landmarks in aligned slices. Three single slices from levels that were 4, 7 and 10 mm caudal to the alignment slice were selected for detailed analysis of velopharyngeal and oropharyngeal airway dimensions and pharyngeal wall tissue displacement and strain. These levels were defined for this study as the rostral, mid- and caudal pharyngeal levels. The velopharynx was defined as the airway dorsal to the soft palate, and the oropharynx, as the airway ventral to the soft palate.

We used NIH-Image (v. 1.61/ppc, US National Institutes of Health http://rsb.info.gov/nih-image/) to measure velopharyngeal and oropharyngeal cross-sectional area (CSA) in axial slices at each pharyngeal level (rostral, mid- and caudal). Lateral and anteroposterior (AP) dimensions were calculated as the major and minor axes of the best fitting ellipse for the airway inscribed from the NIH-Image CSA measurements. The term AP dimension will be used in this report to identify the airway dimension in the ventro-dorsal direction, i.e. perpendicular to the lateral dimension.

Analysis of tissue motion (displacement and strain) in axial slices was performed using the SPAMMVU analysis program. First, an optical flow subprogram was used to estimate pixel displacements between images obtained before and during stimulation (Dougherty et al. 1999). The optical flow method, originally developed for military use for the extraction of motion information from image sequences, was adapted to track the local deformation in tagged pharyngeal airway images and makes use of the SPAMM lines that serve as fiducial markers. The optical flow method is a coarse-to-fine model-based motion-estimation technique for estimating first, a global parametric transformation, and then local deformations (Dougherty et al. 2003). Thus, it computes the flow field that describes the warping of an image of one phase (before stimulation) into alignment with the next (during stimulation). This optical flow software has been adapted to overcome the requirement of constant pixel intensity in standard optical flow methods by pre-processing the input images to reduce any intensity bias which results from the reduction in stripe contrast due to the relaxation of the magnetization of the spins within the tagged area throughout the imaging cycle (Dougherty et al. 1999).

The analysis was performed on a selected rectangular region of interest (1.4 cm × 1.2 cm) surrounding the airways. Once the pixel displacements were determined using the optical flow method, two-dimensional strain analysis using SPAMMVU was performed. While the optical flow method computes the displacement of every pixel in the region of interest, the SPAMMVU program utilizes only a subset of these data (two points per 2.0 mm SPAMM line width) to perform the finite element analysis.

Figure 1 shows a representative caudal axial slice before stimulation (left panel) and during stimulation (middle panel). The third panel (right) shows the before stimulation image (as in left panel) with the displacement data represented as arrows overlaid on the image. These arrows represent the displacement vectors determined from the optical flow analysis. The base of each arrow is located at the initial point of the pixel (i.e. pixel locus before stimulation), and the arrowhead indicates the locus where that pixel was displaced during stimulation. The length and direction of the arrows are equal to the tissue displacements during stimulation. In a few cases there were errors due to aliasing (caused by tissue movements that were greater than a single SPAMM line width) so the displacements were determined manually, using identifiable local tissue landmarks.

Figure 1. SPAMM tracking of pharyngeal tissues during medial hypoglossus stimulation.

Figure 1

Open in a new tab

Caudal level axial images from one rat: before stimulation (left panel), during stimulation (middle panel) and before stimulation image (right panel) with overlay of arrows that show the tissue movements as vectors with base of arrow at the locus before stimulation and the arrowhead pointing to where that pixel was displaced during stimulation. Thus, the displacement data are represented by the arrows whose length and direction are equal to the pixel displacement due to stimulation. The left image is annotated to show: the dorsal (brain) and ventral (tongue) structures and outlined airways are noted as ‘VP’ for velopharynx and ‘OP’ for the oropharynx. Note in the middle panel that enlargement of the oropharynx with stimulation was associated with ventral movement of the SPAMM gridlines in the ventral pharyngeal wall.

A triangulation algorithm inscribed a set of triangles onto the initial points and tracked the transformation of the triangle vertices to the second image. Figure 2 is a schematic diagram that shows an idealized triangle during stimulation. The Results report the displacement of the triangle centroids (not the individual pixel displacements). Thus, displacement measurements are the vectorial components measured for the centroid displacement in the medial–lateral (Fig. 2: x2–x1) and ventral–dorsal (Fig. 2: y2–y1) directions. The tensor that represents the shape change of the triangle is independent of rigid body displacement and can be decomposed into the components of strain and rotation (Axel et al. 1992; Scott et al. 1999). A built-in eigenvalue analysis function of SPAMMVU computes the maximum and minimum principal strains that are called λ1 and λ2, respectively. These values are the orthogonal eigenvectors of the strain tensor and represent the fractional change in length (1 ± (ΔL/Lo) where ΔL is positive for stretch (λ1) and negative for compression (λ2)) of the axes of the unit circle of triangle A that is transformed to an ellipse in triangle B after stimulation (see Fig. 2). SPAMMVU software calculates: the principal major (λ1) and minor (λ2) strains, the direction angle (β) of λ1 relative to the origin, and the angle of rigid body rotation (α) that represents the pure rotation of the triangle excluding shape-related changes (α shown in C inset in Fig. 2).

Figure 2. Schematic representation of tissue displacement and strain.

Figure 2

Open in a new tab

Idealized triangle before and after transformation due to tissue motion. SPAMMVU software analysis of the image data results in the several outcome measures that quantify tissue motion: displacement in the medial–lateral and ventral–dorsal directions, λ1 (stretch), λ2 (compression), β (direction of the λ1 eigenvector) and α (rigid body rotation). (See text for detailed explanation.) This schematic is representative, although not to scale, of tissue displacement in the lateral sectors during medial hypoglossal nerve stimulation. Displacement is shown as the translation of the triangle centroid from (A) to (B). λ1 and λ2 are the orthogonal eigenvectors of the strain tensor and represent the fractional change in length (1 ± (ΔL)/Lo where ΔL is positive for stretch (λ1) and negative for compression (λ2)) of the axes of the unit circle inscribed in triangle A, that would be transformed to an ellipse (triangle B) after stimulation. β angle is the direction of the principal strain (λ1) in a Cartesian coordinate system with 0,0 at the centroid of velopharyngeal airway (the velopharyngeal airway centroid is determined during the SPAMMVU processing). α angle (inset C) indicates the rigid body rotation of the triangle (shown for this example as counter clockwise) that does not include deformation due to shape change. The displacement, rotation, principal strain and directions are derived from tracking the triangle verticies using 2D strain tensor analyses (Axel et al. 1992).

Quantifying tissue motion in specific regions of the pharyngeal wall

To assess tissue motion in specific regions of the pharyngeal wall in the axial images, we selected bilateral square sectors in the lateral and ventral pharyngeal walls. Each sector consisted of eight triangles formed from the points determined by the optical flow methods. The right and left lateral wall sectors contained the points closest to the lateral edges of the oropharyngeal airway. The right and left ventral sectors were adjacent to the mid-sagittal line and included the points nearest to the ventral edge of the oropharyngeal airway. A fifth sector, located in the brain tissue, was selected to serve as a control that theoretically would not show any tissue motion due to stimulation of the pharyngeal muscles. Displacement and strain were measured in the control sectors and these values were used for statistical comparisons with the lateral and ventral pharyngeal wall sectors. Use of an image plane control sector (as opposed to a zero effect unstimulated control value) reduced the chance of over-estimating positive results as noise, errant motion in the images, or optical flow analysis errors would be common to both target and control sectors, thus limiting any bias.

The sectors were readily identified in a repeatable manner (by a trained operator) in the images after triangulation analysis. The sectors in a representative axial image of the caudal pharynx are highlighted in Fig. 3. The averaged strain and displacement data for the eight triangles of each sector were computed. Given the overall symmetry of the measured values, we reduced the sector comparisons from five to three by combining and averaging the data from the two lateral sectors and the two ventral sectors. Medial–lateral displacement results for both the lateral and ventral sectors were combined by multiplying the left sided medial–lateral displacement data by −1, and averaging these with the results on the contralateral side. The ventral–dorsal displacements did not need transformation before averaging. Lateral and ventral sector β and α angles on the left side were converted to right-sided values by the formula: right sided angle = 180 deg – left sided angle. Simple averaging and analysis of the angles was facilitated using angles in radians that were later converted to degrees. The λ values did not require any mirror image conversion before averaging.

Figure 3. Triangulation and pharyngeal airway sector motion.

Figure 3

Open in a new tab

Caudal axial slice processed through triangulation algorithm. The same representative slice from Fig. 1 is displayed here but the point displacements derived by optical flow methods have been assigned to a set of triangles (for use in SPAMMVU eigenvector analyses). The images are annotated to show, velopharynx ‘VP’ and oropharynx ‘OP’ and the ventral and dorsal directions. There are highlighted in each panel, four sectors, each containing eight triangles that were selected bilaterally in the lateral and ventral pharyngeal walls and a fifth sector, located dorsal to the airways in the brain tissue, that served as a control (see Methods). Note the displacement (ventral direction) and shape changes in triangles from the regular pattern prior to stimulation (left panel) to the distorted pattern (right panel) during medial hypoglossal nerve stimulation.

MRI mid-sagittal slices were obtained in all 11 rats. The region of interest analysed in these images was the portion of the tongue ventral to the oropharynx over the length of airway encompassed by the axial slices. As the position of the rat with respect to the magnet was not altered between axial and sagittal acquisitions, the orthogonal alignment feature of the SPAMMVU package was used to identify the relevant axial slices to obtain the region of interest on the sagittal images. Displacement was measured in the rostral–caudal and ventral–dorsal direction of all triangles in the region of interest to assess possible through-plane motion (i.e. rostral–caudal displacement that would be orthogonal to the axial plane where the pharyngeal wall tissues sectors were analysed).

Statistical analysis

We used a mixed model two-way ANOVA on repeated measures (SAS Institute, Cary) (n = 11 rats) to test the effect of stimulation (no stimulation versus stimulation) and pharyngeal level (rostral, mid- and caudal) on airway dimension variables (CSA, AP and lateral dimensions) in both the oropharynx and velopharynx (Littell et al. 1996; Singer, 1998). The ANOVA model was used to test the effect of different levels of pharyngeal region (rostral, mid- and caudal) and pharyngeal wall sectors (ventral, lateral and control) on the tissue displacement and strain variables (ventral–dorsal displacement, medial–lateral displacement, λ1, λ2, λ1 direction angle (β) and ‘rigid-body’ rotation angle (α)). Examination of second-order interactions tested whether pharyngeal level and sector had combined or independent effects on: the tissue displacement (ventral–dorsal or medial–lateral), strain and strain angle variables (λ1, λ2, β and α). Significance for post hoc pair-wise comparisons were evaluated using Tukey-Kramer adjusted P values where significance was assumed for P < 0.05. Post hoc comparisons for subset of specific dimensional parameters were evaluated assuming a limited number of planned contrasts, i.e. stimulated versus non-stimulated in each of three individual regions. In these cases, a Bonferroni adjustment (P < 0.05/3 or 0.017) was used to compare the differences of the least squared means for that contrast or comparison (Winer et al. 1991).

The effect of stimulation on airway shape was analysed by examining the change in elliptical ratio defined as lateral dimension/AP dimension. An elliptical ratio of unity denotes a perfect circle whereas, a ratio smaller than unity represents a prolate ellipsoid (long axis in AP dimension) and a ratio greater than unity represents an oblate ellipsoid (long axes in lateral dimension). If, for example, an airway with elliptical ratio greater than unity becomes more circular during stimulation, then the resulting elliptical ratio would be smaller, i.e. closer to unity. Two-way ANOVA was used to test the effect of stimulation and pharyngeal level on elliptical ratio in the velopharynx and oropharynx. Post hoc analysis of the elliptical ratio was similar to that for the other dimensions.

Results

Figure 4 shows CSA, AP and lateral measurements in the oropharynx (A–C) and velopharynx (D–F) before and during bilateral stimulation. The mixed model ANOVA showed that stimulation significantly increased CSA, and lateral and AP dimensions (P < 0.001) when values in the rostral, mid- and caudal pharynx were combined, and significantly increased these parameters with Bonferroni adjustment, at specific regional levels. For CSA and AP dimension in the oropharynx, and CSA, and lateral and AP dimensions in the velopharynx, there were significant differences among regions when values were averaged over stimulated and non-stimulated conditions (all P < 0.0143). The inequalities between rostral, mid- and caudal regions are noted on each panel in Fig. 4 where significance met the Bonferroni criteria, P < 0.017. There were no significant interactions between stimulation and pharyngeal region for CSA, lateral dimension or AP dimension in either the oropharynx or velopharynx (all P = 0.086). Thus, for a given airway parameter, stimulation had a similar effect across pharyngeal regions.

Figure 4. Oropharyngeal and velopharyngeal airway dimensions at rostral, mid- and caudal pharyngeal regions.

Figure 4

Open in a new tab

Mean (± s.e.m.) CSA, lateral and AP measurements in the oropharynx (A–C) and velopharynx (D–F) in 11 rats. Filled circles and triangles (• ▴) represent, respectively, oropharyngeal and velopharyngeal measurements without stimulation and open circles and triangles (○ Δ) represent, respectively, oropharyngeal and velopharyngeal measurements with stimulation. *Stimulated values are significantly greater than unstimulated values at the same pharyngeal region (P < 0.017 with Bonferroni adjustment). Averaged non-stimulated and stimulated values were significantly different by region as noted in each panel. For example in (A): R, M < C (P < 0.017) to indicate that the mean caudal (C) CSA of non-stimulated and stimulated values were greater than both mid- (M) and rostral (R) values.

Before stimulation, the oropharyngeal and velopharyngeal airways had an oblate ellipsoid shape (ratio was greater than unity), and in any given region, the oropharynx had a greater elliptical ratio than the velopharynx (P < 0.001). The analysis of the airways' elliptical shape due to stimulation is summarized in Table 1. When averaged over all regions, medial hypoglossal nerve stimulation significantly reduced the elliptical ratio in the oropharyngeal airway (P < 0.0015). The oblate ellipsoid shape of the oropharyngeal airway without stimulation became more circular during stimulation but remained ellipsoid with the longest axis in the lateral dimension. The amount of shape change with stimulation did not depend on which region was stimulated (interaction term, stimulation × region, P = 0.51) although there were significant regional differences in the ratio when averaged over non-stimulated and stimulated conditions (rostral and mid- > caudal, P < 0.017). In the velopharynx, stimulation had no significant effect on the velopharyngeal elliptical ratios (P = 0.079). However, regional differences in the elliptical ratio did exist, such that elliptical ratios in the rostral and mid-velopharynx were significantly greater than the caudal values (P < 0.0001). Although the stimulation did not significantly effect the velopharyngeal elliptical ratio, the interaction of stimulation and region showed P < 0.10 and this, according to Winer et al. (1991) is sufficient for post hoc tests. Examination of the differences of least squares means revealed that stimulation significantly decreased the elliptical ratio in the caudal velopharynx (P < 0.006; Bonferroni significant at < 0.017). Thus, in the caudal velopharynx as well as the oropharynx (averaged over all regions) stimulation changed the elliptical airways towards a more circular shape.

Table 1.

Oropharyngeal and velopharyngeal airway shape changes due to stimulation

Rostral Mid Caudal
Oropharynx
 No stimulation 4.71 ± 0.59 4.99 ± 0.72 3.80 ± 0.31
 Stimulation 4.06 ± 0.56 3.34 ± 0.32 2.32 ± 0.27
Velopharynx
 No stimulation 1.38 ± 0.10 1.45 ± 0.07 2.08 ± 0.19
 Stimulation 1.46 ± 0.13 1.33 ± 0.07 1.65 ± 0.10
Open in a new tab

Data are means ± s.e.m. In the oropharynx, stimulation caused a significant reduction in the (dimensionless) elliptical ratio when averaged over all regions (P < 0.002), indicating that the airway became more circular. In the velopharynx, stimulation had no significant effect on elliptical ratio (P = 0.11). When both stimulated and non-stimulated values were averaged, in the oropharynx, the rostral elliptical ratio values were greater than the caudal values (P < 0.009). However, in the oropharynx, regional differences in shape did not affect the decrease in elliptical ratio due to stimulation (interaction term, stimulation × region, P = 0.55). In the velopharynx, stimulation did not cause a significant change in elliptical ratio (P = 0.11); however, when regional velopharyngeal airway elliptical ratios were compared, the caudal region had a greater mean elliptical ratio than both mid- and rostral regions (P < 0.0001).

The effects of nerve stimulation on tissue displacement in the ventral, lateral and dorsal (control) sectors are shown for one rat in the mid-pharynx in Fig. 3 (right panel) and for all rats at all levels in Fig. 5. The predominant effect of stimulation was to move the ventral and lateral pharyngeal walls in a ventral direction with the greatest effect in the ventral pharyngeal walls (Fig. 5, right panel). Although the ventral wall sector displacements in the ventral direction were greatest at any level, lateral wall displacements in the ventral direction were also greater than control in the mid- and caudal pharyngeal levels (Tukey-Kramer, P < 0.0001). The magnitude of ventral displacement in a given wall sector was always smaller than the increase in oropharyngeal airway AP dimension (compare Fig. 4C to Fig. 5, right panel). There was minimal or near zero medial–lateral and ventral–dorsal displacements for the control sector (dorsal brain region). This result indicated that there was little overall head movement during nerve stimulation.

Figure 5. Pharyngeal wall tissue displacement.

Figure 5

Open in a new tab

Left and right panels show, respectively, the mean (± s.e.m.) tissue displacements in medial–lateral and ventral–caudal directions in the ventral (□), lateral (⊠), and dorsal (▪) sectors in the rostral, mid- and caudal pharynx. In each plot, displacement (ordinate) represents movement of a right-sided sector relative to the centroid of the velopharynx such that, positive medial–lateral displacements are directed laterally, and positive dorsal–ventral displacements are directed ventrally. Note in the left panel, the lateral sectors in the mid- and caudal regions showed significant displacement (less than control) in the medial direction (negative values) (P < 0.0001). Lateral sector movement was greater (in the negative diretion) than both ventral and control (*). Comparison of lateral sector displacement among regions showed that the caudal region (†) was significantly less that that in the mid- region (††) (P < 0.016). In the right panel, ventrally directed displacement in both ventral and lateral sectors was significantly greater than that in control (all P < 0.003). When compared at a given region, ventral sector displacements (**) were greater than those of the lateral sectors (*). Compared among regions, ventral displacement of ventral sectors in the mid- and caudal regions (††) were significantly greater than ventral displacement of ventral sectors in the rostral region (†) (P < 0.03). Note also, that ventral displacement of lateral sectors was significantly different by region such that: rostral < mid- (†) < caudal (††) (P < 0.03).

Significant ventral displacement of the lateral walls (0.23 ± 0.04 and 0.40 ± 0.05 mm, respectively, in the mid- and caudal regions; Tukey-Kramer, P < 0.0001) was accompanied by small, but significant displacement in the medial direction (Fig. 5, left panel) at 0.11 ± 0.01 and 0.19 ± 0.02 mm, respectively (Tukey-Kramer, P < 0.0001). This is consistent with the direction of the displacement arrows in the lateral pharyngeal wall in Fig. 3 (right panel) that point in ventral–medial direction. These findings were further confirmed by visually reviewing the tissue displacements in all of the images analysed. None showed outward displacement of lateral wall sectors with stimulation despite simultaneous increases in the airway's lateral dimensions.

Figure 6 shows λ1 and λ2 values compared by sector and pharyngeal region. λ1 (tissue stretch) in the lateral sector was significantly greater than that in the control sector in both the mid- and caudal regions (lateral versus control: mid-, P < 0.05; lateral versus control: caudal, P < 0.0001). Tissue stretch in lateral sectors was 24 and 30% greater than the initial length (Lo) in the mid- and caudal regions, respectively. λ2 (tissue compression) values were significantly different in the order: ventral < lateral < control (Tukey-Kramer, P < 0.05) in all regions, and there were significant differences when comparing the average λ2 values across regions (mid- > caudal, Tukey-Kramer, P < 0.013). For λ2 values there were no significant interactions between sector (ventral, lateral and control) and pharyngeal region (rostral, mid- and caudal). Stimulation therefore was associated with significant stretch of tissues in the lateral pharyngeal wall sectors in the mid- and caudal pharyngeal regions and significant compression in both the ventral and lateral pharyngeal wall sectors in all regions.

Figure 6. Stretch and compression in pharyngeal wall sectors.

Figure 6

Open in a new tab

In upper panel are mean (± s.e.m.) λ1 values (stretch) for each sector in the rostral, mid- and caudal pharynx. Lower panel shows the mean λ2 (± s.e.m.) values (compression) in the same format. λ1 values are shown as the fractional increase over Lo, and λ2 (compression) is the fractional reduction in Lo (Lo = unit value of initial length). Values noted (*, **) significantly different from control with (*) values > (**) values. For λ1 mid- region, lateral > control (P < 0.05) and in the caudal region, lateral > control (P < 0.0001). λ2 (tissue compression) values were significantly different and in the order: ventral < lateral < control (Tukey-Kramer, P < 0.05) in all regions, and there was a significant differences when comparing the average λ2 values across regions (mid- > low, Tukey-Kramer, P < 0.013). The control values of λ1 and λ2 strains measured in the dorsal brain sector deviate slightly from the ideal unit value of Lo = 1 and this was assumed to be due to random noise in the images.

The direction of the maximum stretch for each sector in each region is shown as the β angle in Table 2A. β angles in the lateral and ventral sectors were significantly greater than control (P < 0.0001). However, the control values had a large variance. As the measured strain (λ1 and λ2) in the control sectors was only minimally different from unity, (i.e. negligible strain; see Fig. 6), we compared the β angles between lateral and ventral sectors across pharyngeal levels by two-way ANOVA. Thus, in the mid- and caudal regions, where λ1 was significantly greater than control, the β angles in the ventral sectors were nearly perpendicular to the airway in a ventral direction (73.0 ± 5.4 deg ≤β≤ 84.2 ± 6.2 deg). In the lateral sectors for the mid- and caudal regions, the β angles were directed in a more ventral–lateral direction (43.1 ± 7.4 deg ≤β≤ 53.4 ± 7.4 deg). The direction of maximum tissue compression is by definition perpendicular to the β angle. The α angle results in Table 2B indicate that, compared to control, there was no significant tissue rotation with stimulation in either ventral or lateral sectors in all regions. The mean absolute maximum rotation in any sector was not greater than 1.67 ± 1.27 deg.

Table 2.

β and α angles in pharyngeal wall tissue sectors

Sector Rostral Mid Caudal
A. β angles (deg)
 Ventral  79.4 ± 6.4**  73.0 ± 5.4**  84.2 ± 6.2**
 Lateral  29.5 ± 6.1*  43.1 ± 7.4*  53.4 ± 7.4*
 Control 240.6 ± 41.2 227.9 ± 40.3 216.5 ± 46.4 
B. α angles (deg)
 Ventral  1.6 ± 1.4 −0.3 ± 1.0 1.7 ± 1.3
 Lateral −0.6 ± 0.2  0.2 ± 0.8 0.7 ± 0.5
 Control −0.2 ± 0.2  0.2 ± 0.2 0.4 ± 0.6
Open in a new tab

A, mean β angles (1± s.e.m.) calculated for each sector in each region. Both ventral and lateral sector angles were significantly different than control (P < 0.0001); and ventral β angles marked

(**)

were larger than lateral angles marked

(*)

(P < 0.0001; two-way ANOVA excluding control sector β angles, see text. There were no significant differences in β angles between regions: rostral, mid- or caudal. B, mean α angles (± s.e.m.) in each sector in each region. α angles are referenced from the triangle centroids from each sector and, using the Cartesian coordinate system, a positive angle represents a counter-clockwise rotation. There were no significant differences in α angles across sectors (lateral, ventral and control) or pharyngeal levels (rostral, mid- and caudal).

Figure 7 shows a mid-sagittal image in a representative rat before and during stimulation. Examination of the mid-sagittal images suggested that stimulation caused tongue movement orthogonal to the axial plane. The region of interest identified by the triangles in Fig. 7 was chosen to include the region where measurements were obtained in the axial images. Overall, the data in Table 3 show that the mean maximum absolute value of movement for any single rat, along the rostral–caudal axis was 0.78 ± 0.03 mm while the mean absolute movement along the rostral–caudal axis of all rats was only 0.12 ± 0.06 mm. These values were less than the slice thickness of 1.0 mm in the axial direction and therefore support the fact that the analysis of the axial images was not affected by through plane pixel motion. The mean sagittal, ventral–dorsal displacement of the triangles in the sagittal region of interest (0.88 ± 0.03 mm; Table 3) was similar to the mean ventral–dorsal displacement of the ventral sectors measured in the axial slices (0.76 ± 0.08 mm; Fig. 5, right panel). Therefore, measurement of tissue movement in different orthogonal MRI slices gave similar results.

Figure 7. Effect of stimulation in mid-sagittal images.

Figure 7

Open in a new tab

A representative mid-sagittal SPGR image with SPAMM (from the same rat as illustrated in Figs 1 and 3) before stimulation (bottom image) and during stimulation (top image). The top image is annotated to show the ventral and dorsal regions of the supine rat, the tongue and soft palate (SP). The triangulated sector in the tongue region was selected to encompass the rostral, mid- and caudal axial slices that were obtained in the transverse axial images. Displacement of the triangles in the rostral to caudal direction indicates that there was some tongue motion orthogonal to the axial plane. There was also displacement of the triangles in the dorsal to ventral direction that was of similar magnitude and direction as that measured in the ventral pharyngeal wall sectors in axial plane slices (see Fig. 5). A summary of the rostral–caudal and ventral–dorsal displacements measured in the sagittal slices is given in Table 2.

Table 3.

Mean sagittal plan displacements (mm) in region of interest

rat ID Rostral–caudal s.e.m. Ventral–dorsal s.e.m.
1  0.26 0.07 0.82 0.02
2  0.31 0.08 0.40 0.05
3 −0.06 0.02 0.21 0.03
4  0.36 0.08 0.46 0.02
5  0.29 0.10 0.86 0.02
6  0.14 0.08 0.81 0.04
7 −0.33 0.07 1.54 0.03
8 −0.53 0.04 1.24 0.03
9 −0.69 0.03 1.24 0.03
10 −0.78 0.03 0.78 0.03
11 −0.27 0.04 1.36 0.03
Averages −0.12 0.06 0.88 0.03
Open in a new tab

Mean rostral–caudal and ventral–dorsal displacements of tissues in the mid-sagittal plane. The average rostral–caudal displacement was −0.12 ± 0.06 mm compared to 1.0 mm thickness of axial slices. The average ventral–dorsal displacement was 0.88 ± 0.03 and this value was comparable to the ventral sector displacement (4.9–9.5 mm) that was from a tissue volume in the same region of interest measured in axial slices. Note that rostral–caudal displacement is orthogonal to axial images; positive (+) direction is rostral, and ventral–dorsal displacement is in same direction as ventral–dorsal in axial images; positive (+) direction is ventral.

Discussion

In this study, we have introduced MRI with SPAMM tagging to measure tissue movements in the pharyngeal wall during stimulation of the medial hypoglossal nerve in rats. We addressed two important questions. (1) What is the effect of medial hypoglossal nerve stimulation on the motion (displacement and strain) of the pharyngeal wall tissues? (2) How do the stimulation-induced changes in airway area and shape in the rostral, mid- and caudal pharyngeal regions relate to the simultaneous displacement and strain of the ventral and lateral pharyngeal wall tissues? Activation of the protrudor and intrinsic tongue muscles with medial hypoglossal stimulation caused a significant increase in oropharyngeal and velopharyngeal CSA in the rostral, mid- and caudal pharynx. The increase in the airway's AP dimension in the oropharynx was proportionally greater than the increase in lateral dimension, as evidenced by a significant decrease of the oropharyngeal elliptical ratio with stimulation. The increase in airway size (and change to a more circular shape in the oropharynx) was associated with significant ventral displacement of tissues in the ventral pharyngeal wall sectors at all three airway levels, and ventral displacement of lateral wall sectors in the mid- and caudal levels. Analysis of the strain (independent of tissue displacement) showed that the principal stretch (λ1) in the lateral pharyngeal wall sectors the in the mid- and caudal pharyngeal regions was significantly greater than control sectors. In addition, there was in all regions, a significant compression (λ2) of both ventral and lateral pharyngeal wall tissues. The strain analysis implies that there was some resultant force (stress) that was pulling the lateral tissues in a ventral–lateral direction causing a significant degree of stretch in those tissues. Significant compression of the both the lateral and ventral tissues may indicate either active contraction in those tissues or compression that resulted from other active contractile tissues being stimulated by the medial hypoglossal nerve. Overall, the results reveal that the increase in airway size and change in shape from oblate ellipsoid to a more circular shape, resulting from stimulation of the medial branch of the hypoglossal nerve, is predominantly due to ventral motion of the ventral and lateral pharyngeal walls.

In response to stimulation, the lateral pharyngeal wall sector was displaced in a ventral or ventral–medial direction but stretched in a ventral–lateral direction. This lateral pharyngeal wall tissue motion was associated with a significant increase in lateral oropharyngeal airway dimension. Lateral displacement of the lateral pharyngeal walls with stimulation was never observed. Therefore, changes in airway size (AP or lateral) with hypoglossal nerve stimulation cannot be explained simply on that basis of radial traction (i.e. a dilating force perpendicular to the tangent of the airway circumference). Instead, airway dilatation caused by stimulation of the medial branch of the hypoglossus appears to involve complex mechanical relationships, whereby ventral displacement and ventral–lateral stretch of the pharyngeal wall tissues causes significant increases in both the anteroposterior and lateral airway dimension.

Given the importance of state-related changes in pharyngeal muscle activity in the pathogenesis of obstructive sleep apnoea, numerous previous studies have examined the mechanical effects of pharyngeal muscle activation on airway function using a variety of techniques (Brouillette & Thach, 1980; Strohl et al. 1987; Hida et al. 1995; Eisele et al. 1997; Ilomaki et al. 1997; Fuller et al. 1998; Isono et al. 1999). Physiological studies have determined the effects of pharyngeal muscle activation on airway collapsibility by measuring pressure–volume (area) relationships in an isolated sealed upper airway under static conditions and critical airway pressure under dynamic conditions (Strohl et al. 1987; Schwartz et al. 1993, 1998; Hida et al. 1995; Fuller et al. 1999). These studies demonstrate that activation of the tongue protrudor muscles dilates and stiffens the pharyngeal airway. The techniques employed provide a global assessment of pharyngeal airway mechanics, but are limited in their ability to localize the effects to a particular region of the airway. Imaging of the pharyngeal airway with fibre optics, CT or MRI can localize effects to specific regions of the airway (Suratt et al. 1983; Shepard & Thawley, 1990; Ryan & Love, 1996). For example, recent fibre optic studies from this laboratory in animals have shown that selective stimulation of pharyngeal airway muscles has very different effects on different regions of the pharyngeal airway (Kuna, 2001, Kuna, 2004; Kuna & Brennick, 2002). Although fibre optic imaging can evaluate changes in airway size and shape, this technique provides no information about pharyngeal wall tissues. CT and MRI have been used to evaluate the pharyngeal airway and its surrounding soft tissue volumes in humans, and investigators report that patients with obstructive sleep apnoea have a reduced pharyngeal airway volume and increased volume of soft tissue structures in the airway wall (Horner et al. 1989a, 1989b; Shelton et al. 1993; Schwab et al. 1996, 2003). However, like fibre optic imaging, conventional CT and MRI are unable to directly evaluate pharyngeal wall tissue motion. The use of MRI with SPAMM allows, for the first time, the ability to track tissue motion in the pharyngeal wall and explore how this motion translates into changes in airway size and shape.

The changes in airway CSA with medial hypoglossal stimulation in the current study support results from a previous MRI study in rats (Brennick et al. 2001). In agreement with those previous findings, the current study found that medial hypoglossal branch stimulation caused significant oropharyngeal and velopharyngeal dilatation. The previous study did not include AP or lateral dimensional analysis and the analysis of soft tissue motion was limited to a single linear measurement of tongue displacement in the ventral direction. The current study extends these findings by showing that increases in AP and lateral airway dimensions with stimulation are associated with displacement and strain in discrete sectors of the pharyngeal walls. The ability to quantify tissue motion in the pharyngeal wall represents a powerful new technique to assess pharyngeal mechanics.

Although MRI with SPAMM has been thoroughly tested and detailed in previous studies of cardiac and tongue motion (Axel et al. 1992; Young et al. 1993; Niitsu et al. 1994; Marcus et al. 1997; Napadow et al. 1999; Scott et al. 1999; Yuan et al. 2000; Stone et al. 2001; Yeon et al. 2001), we will address some important technical aspects of this study as they relate to our novel application of gated MRI with SPAMM using nerve stimulation to study pharyngeal airway mechanics. The degree of spatial and temporal resolution in MRI is subject to the limits of the equipment. We controlled temporal resolution to a large degree by using a stimulus-gated protocol. By this method, nerve stimulation was initiated at a precise time between acquisition of the two images. This stimulus-gated protocol is based on standard MRI gated imaging techniques and offers the advantage of being able to adjust the cycle time with an external stimulator. This contrasts with cardiac gated imaging where heart rate, although fairly constant, is not under direct experimental control (Yuan et al. 2000). Control of cycle time allowed us to increase spatial resolution because, in the protocol used in this study, longer cycle time increases TR and this improves the image quality for SPGR imaging. Overall, spatial resolution is a result of many factors in MRI including magnet field strength, gradient capabilities, technical specifics of the RF imaging coil and the MRI protocol design. Whenever possible, we optimized variables to obtain the best possible image quality.

We chose to employ a SPGR echo sequence as this would allow two or more images to be acquired following the deposition of SPAMM lines. Although spin echo has been used with SPAMM applications (Crespigny et al. 1991), the drawback is that the total imaging process is much longer as the TR needed for spin echo sequence is of the order of 1 s or more compared to the 200 ms or less needed for the SPGR sequence. Thus, we tried to maximize the spatial resolution in the SPGR imaging by maximizing TR and minimizing TE (we used 0.2 ms, which was approximately the limit of our equipment). We also found that it was preferable to extend the cycle time from 480 ms to 630 ms. There was no loss of SPAMM line contrast with this adjustment, and the longer cycle time reduced the stimulated muscle duty cycle.

SPAMMVU software was designed for 2-dimensional (2D) motion and strain analysis in the heart, and the significant structural differences between the heart and pharynx potentially could affect the analysis and interpretation of the results in pharyngeal tissue. In general, cardiac tissue is relatively homogeneous with alignment of muscle fibres in an overlaying pattern in order to produce a single repetitive function (Noordegraaf, 1978). During systole, cardiac ventricular muscle contracts in a unified manner or as a syncytium and thus, when viewed in transverse short axis slices, the cardiac wall stretches in the radial direction and compresses tissue in the circumferential direction (Axel et al. 1992; Scott et al. 1999). In addition, the ventricles have clearly defined inner and outer surfaces. In contrast, the pharynx is a much more heterogeneous structure. The pharyngeal skeletal muscles are anatomically arranged in complex and varied patterns, and they are not generally activated in unison (Bartlett, 1986). The pharyngeal walls are not clearly bounded on the non-airway side and can be quite thick in some regions especially the ventral part that is primarily composed of the tongue. To compensate for these anatomical differences between the heart and pharynx, we developed a method utilizing the SPAMMVU software to examine tissue movements in discrete, selected regions of the pharyngeal wall. In addition, while the geometry of the heart lends itself to a cylindrical coordinate system, we used a Cartesian coordinate system to represent pharyngeal tissue displacement and strain. These changes did not affect the theoretical basis for strain calculations (Axel et al. 1992). However, they did allow us to extend the SPAMMVU software to the unique anatomical characteristics of the pharynx. Thus, we were able to adapt the 2D tissue analysis software SPAMMVU to the pharynx and develope a method that quantifies local deformation and displacement of pharyngeal tissues in any region of interest.

We used an optical flow method to determine the estimated pixel displacements from the unstimulated to stimulated images (Dougherty et al. 1999). A validation study has shown that this optical flow method used with MRI of a mechanically rotated test phantom was within 4% of known values and within 6.7% of results obtained with a semiautomated, active contours model for SPAMM analysis (Dougherty et al. 1999). In some axial images there was a large movement from unstimulated image to stimulated image and this condition where the movement was greater than 1/2 SPAMM gridline separation, occurred in some instances and required manual validation or modification of points. Although the power requirements of the pulse sequence that produces the SPAMM lines limit the reduction of SPAMM line width and gridline separation, we were able to work within these requirements to obtain a SPAMM line width of approximately 0.27 mm or 2 pixels wide and a gridline separation of 2.0 mm or 14.6 pixels. We felt that this combination was optimal for imaging the pharynx of the rat.

We accounted for through plane motion using methods described by Marcus et al. (1997). In their MRI with SPAMM study of myocardial function, they used 2D strain analysis to quantify myocardial strain and accounted for through plane motion by acquiring complementary image series in orthogonal planes. They reasoned that where significant through plane motion was evident (approximately one slice thickness) then 2D results in one plane should be confirmed by analysis on the orthogonal plane. In the current study, we found that tissue displacement in the rostral–caudal direction was less than the slice thickness. In addition, we found that the mean ventral tissue displacements on the sagittal images were of the same magnitude as those in the axial slices from a comparable although more narrowly defined region, i.e. the ventral sectors adjacent to the oropharynx. Thus, although it is possible that through plane motion may have been a source of some measurement error, we believe that the overall 2D strain and displacements measurements accurately represent pharyngeal tissue movements.

In summary, we have introduced a new method to examine pharyngeal mechanics using MRI with SPAMM non-invasive tissue tagging in rats. In this study bilateral stimulation of the medial hypoglossal nerve caused a significant increase in oropharyngeal and velopharyngeal cross-sectional airway area in the rostral, mid- and caudal pharynx and increased the anteroposterior and lateral dimensions in all three levels of the oropharynx. This was associated with significant ventral displacement of tissues in the ventral and lateral walls. Although the increase in the oropharyngeal airway's lateral dimension was not associated with lateral displacement of the lateral pharyngeal wall tissues, there was significant stretch of lateral wall tissues in a ventral–lateral direction. Thus, the changes in airway size in a particular dimension reflect complex mechanical relationships. Overall, the significant increases in both the AP and lateral airway dimensions caused by stimulation of tongue protrudors involve ventral displacement of pharyngeal walls. The use of MRI with SPAMM to study pharyngeal tissue motion and the resulting changes in airway size may reveal new insights into pharyngeal mechanics and improve our understanding of the pathophysiology and treatment of obstructive sleep apnoea.

Acknowledgments

The authors gratefully acknowledge the technical contributions of Jonathan Palma. We are also grateful to Dr Victor A. Ferrari for his advice on this work. This work was funded in part by NIH HL-27520 and NIH EB-01780.

References

  1. Axel L, Dougherty L. MR imaging of motion with spatial modulation of magnetization. Radiology. 1989;171:841–845. doi: 10.1148/radiology.171.3.2717762. [DOI] [PubMed] [Google Scholar]
  2. Axel L, Goncalves RC, Bloomgarden D. Regional heart wall motion: two dimensional analysis and functional imaging with MR imaging. Radiology. 1992;183:745–750. doi: 10.1148/radiology.183.3.1584931. [DOI] [PubMed] [Google Scholar]
  3. Bartlett D., Jr . Upper Airway Motor Systems. In: Fishman AP, editor. The Respiratory System. Vol. 2. Baltimore: American Physiology Society; 1986. pp. 223–246. [Google Scholar]
  4. Brennick MJ, Ogilvie MD, Margulies SS, Gefter WB, Pack AI. MRI study of regional variations in pharyngeal wall compliance in cats. J Appl Physiol. 1998;85:1884–1897. doi: 10.1152/jappl.1998.85.5.1884. [DOI] [PubMed] [Google Scholar]
  5. Brennick MJ, Trouard TP, Gmitro AF, Fregosi RF. MRI study of pharyngeal airway changes during stimulation of the hypoglossal nerve branches in rats. J Appl Physiol. 2001;90:1373–1384. doi: 10.1152/jappl.2001.90.4.1373. [DOI] [PubMed] [Google Scholar]
  6. Brouillette RT, Thach BT. A neuromuscular mechanism maintaining extrathoracic airway patency. J Appl Physiol. 1979;46:772–779. doi: 10.1152/jappl.1979.46.4.772. [DOI] [PubMed] [Google Scholar]
  7. Brouillette RT, Thach BT. Control of genioglossus muscle inspiratory activity. J Appl Physiol. 1980;49:801–808. doi: 10.1152/jappl.1980.49.5.801. [DOI] [PubMed] [Google Scholar]
  8. Crespigny AJS, Carpenter TA, Hall LD. Cardiac tagging in a rat using a DANTE sequence. Magn Reson Med. 1991;21:151–156. doi: 10.1002/mrm.1910210119. [DOI] [PubMed] [Google Scholar]
  9. Dougherty L, Asmuth JC, Blom AS, Kumar R, Axel L. Validation of an optical flow method for tag displacement estimation. IEEE Trans Med Imaging. 1999;18:359–363. doi: 10.1109/42.768845. [DOI] [PubMed] [Google Scholar]
  10. Dougherty L, Asmuth JC, Gefter WB. Alignment of CT lung volumes with an optical flow method. Acad Radiol. 2003;10:249–254. doi: 10.1016/s1076-6332(03)80098-3. [DOI] [PubMed] [Google Scholar]
  11. Eisele DW, Schwartz AR, Hari A, Thut DC, Smith PL. The effects of selective nerve stimulation on upper airway airflow mechanics. Arch Otolaryngol Head Neck Surg. 1995;121:1361–1364. doi: 10.1001/archotol.1995.01890120021004. [DOI] [PubMed] [Google Scholar]
  12. Eisele DW, Smith PL, Alam DS, Schwartz AR. Direct hypoglossal nerve stimulation in obstructive sleep apnea. Arch Otolaryngol Head Neck Surg. 1997;123:57–61. doi: 10.1001/archotol.1997.01900010067009. [DOI] [PubMed] [Google Scholar]
  13. Fregosi RF, Fuller DD. Respiratory-related control of extrinsic tongue muscle activity. Respir Physiol. 1997;110:295–306. doi: 10.1016/s0034-5687(97)00095-9. [DOI] [PubMed] [Google Scholar]
  14. Fuller D, Mateika JH, Fregosi RF. Co-activation of tongue protrudor and retractor muscles during chemoreceptor stimulation in the rat. J Physiol. 1998;507:265–276. doi: 10.1111/j.1469-7793.1998.265bu.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Fuller DD, Williams PL, Janssen PL, Fregosi RF. Effect of co-activation of tongue protrudor and retractor muscles on tongue movements and pharyngeal airflow mechanics in the rat. J Physiol. 1999;519:601–613. doi: 10.1111/j.1469-7793.1999.0601m.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gotte MJW, van Rossum AC, Marcus JT, Kuijer JPA, Axel L, Visser CA. Recognition of infarct localization by specific changes in intramural myocardial mechanics. Am Heart J. 1999;138:1038–1045. doi: 10.1016/s0002-8703(99)70068-2. [DOI] [PubMed] [Google Scholar]
  17. Hida W, Kurosawa H, Okabe S, Kikuchi Y, Midorikawa J, Chung Y, Takishima T, Shirato K. Hypoglossal nerve stimulation affects the pressure-volume behavior of the upper airway. Am J Respir Crit Care Med. 1995;151:455–460. doi: 10.1164/ajrccm.151.2.7842206. [DOI] [PubMed] [Google Scholar]
  18. Horner RL, Mohiaddin RH, Lowell DG, Shea SA, Burman ED, Longmore DB, Guz A. Sites and sizes of fat deposits around the pharynx in obese patients with obstructive sleep apnoea and weight matched controls. Eur Respir J. 1989a;2:613–622. [PubMed] [Google Scholar]
  19. Horner RL, Shea SA, McIvor J, Guz A. Pharyngeal size and shape during wakefulness and sleep in patients with obstructive sleep apnea. Q J Med. 1989b;72:719–735. [PubMed] [Google Scholar]
  20. Ilomaki J, Baer GA, Karhuketo T, Talonen P, Puhakka H. Pharyngeal patency caused by stimulation of the hypoglossal nerve in anaesthsia-relaxed patients. Acta Otolaryngol Suppl. 1997;529:210–211. doi: 10.3109/00016489709124124. [DOI] [PubMed] [Google Scholar]
  21. Isono S, Remmers JE, Tanaka A, Sho Y, Sato J, Nishino T. Anatomy of pharynx in patients with obstructive sleep apnea and in normal subjects. J Appl Physiol. 1997;82:1319–1326. doi: 10.1152/jappl.1997.82.4.1319. [DOI] [PubMed] [Google Scholar]
  22. Isono S, Shimada A, Tanaka A, Tagaito Y, Utsugi M, Konno A, Nishino T. Efficacy of endoscopic static pressure/area assessment of the passive pharynx in predicting uvulopalatopharyngoplasty outcomes. Laryngoscope. 1999;109:769–774. doi: 10.1097/00005537-199905000-00016. [DOI] [PubMed] [Google Scholar]
  23. Kuna ST. Effects of pharyngeal muscle activation on airway size and configuration. Am J Respir Crit Care Med. 2001;164:1236–1241. doi: 10.1164/ajrccm.164.7.2011030. [DOI] [PubMed] [Google Scholar]
  24. Kuna ST. Regional effects of selective pharyngeal muscle activation on airway shape. Am J Respir Crit Care Med. 2004;169:1063–1069. doi: 10.1164/rccm.200309-1283OC. [DOI] [PubMed] [Google Scholar]
  25. Kuna ST, Brennick MJ. Effects of pharyngeal muscle activation on pressure-area relationships. Am J Respir Crit Care Med. 2002;166:972–977. doi: 10.1164/rccm.200203-214OC. [DOI] [PubMed] [Google Scholar]
  26. Littell RC, Milliken GA, Stroup WW, Wolfinger RD. SAS@ System for Mixed Models. Cary: Sas Institute Inc; 1996. [Google Scholar]
  27. Marcus JT, Gotte MJW, VanRossum AC, Kuijer JPA, Heethaar RM, Axel L, Visser CA. Myocardial function in infarcted and remote regions early after infarction in man: assessment by magnetic resonance tagging and strain analysis. Magn Reson Med. 1997;38:803–810. doi: 10.1002/mrm.1910380517. [DOI] [PubMed] [Google Scholar]
  28. Morrell MJ, Arabi Y, Zahn B, Badr MS. Progressive retropalatal narrowing preceeding obstructive sleep apnea. Am J Respir Crit Care Med. 1998;158:1974–1981. doi: 10.1164/ajrccm.158.6.9712107. [DOI] [PubMed] [Google Scholar]
  29. Napadow VJ, Chen Q, Wedeen VJ, Gilbert RJ. Intramural mechanics of the human tongue in association with physiological deformations. J Biomech. 1999;32:1–12. doi: 10.1016/s0021-9290(98)00109-2. [DOI] [PubMed] [Google Scholar]
  30. Niitsu M, Kumada M, Campeau NG, Niimi S, Riederer SJ, Itai Y. Tongue displacement: visualization with rapid tagged magnetization-prepared MR imaging. Radiology. 1994;191:578–580. doi: 10.1148/radiology.191.2.8153346. [DOI] [PubMed] [Google Scholar]
  31. Noordegraaf A. Circulatory System Dynamics. New York: Academic Press; 1978. The Heart; pp. 198–249. [Google Scholar]
  32. Oliven A, Schnall RP, Pillar G, Gavriely N, Odeh M. Sublingual electrical stimulation of the tongue during wakefulness and sleep. Respir Physiol. 2001;127:217–226. doi: 10.1016/s0034-5687(01)00254-7. [DOI] [PubMed] [Google Scholar]
  33. Redline S, Strohl KP. Influence of upper airway sensory receptors on respiratory muscle activation in humans. J Appl Physiol. 1987;63:368–374. doi: 10.1152/jappl.1987.63.1.368. [DOI] [PubMed] [Google Scholar]
  34. Ryan CF, Love LL. Mechanical properties of the velopharynx in obese patients with obstructive sleep apnea. Am J Respir Crit Care Med. 1996;154:806–812. doi: 10.1164/ajrccm.154.3.8810623. [DOI] [PubMed] [Google Scholar]
  35. Schwab RJ, Gupta KB, Gefter WB, Metzger LJ, Hoffman EA, Pack AI. Upper airway and soft tissue anatomy in normal subjects and patients with sleep disordered breathing. Am J Respir Crit Care Med. 1995;152:1673–1689. doi: 10.1164/ajrccm.152.5.7582313. [DOI] [PubMed] [Google Scholar]
  36. Schwab RJ, Pack AI, Gupta KB, Metzger LJ, Oh E, Getsy JE, Hoffman EA, Gefter WB. Upper airway and soft tissue structural changes induced by CPAP in normal subjects. Am J Respir Crit Care Med. 1996;154:1106–1116. doi: 10.1164/ajrccm.154.4.8887615. [DOI] [PubMed] [Google Scholar]
  37. Schwab RJ, Pairstein M, Pierson R, Mackley A, Hachadoorian R, Arens R, Maislin G, Pack AI. Identification of upper airway anatomic risk factors for obstructive sleep apnea with volumetric magnetic resonance imaging. Am J Respir Crit Care Med. 2003;168:522–530. doi: 10.1164/rccm.200208-866OC. [DOI] [PubMed] [Google Scholar]
  38. Schwartz AR, Eisele DW, Hari A, Testerman R, Erickson D, Smith PL. Electrical stimulation of the lingual musculature in obstructive sleep apnea. J Appl Physiol. 1996;81:643–652. doi: 10.1152/jappl.1996.81.2.643. [DOI] [PubMed] [Google Scholar]
  39. Schwartz AR, O'Donnell CP, Baron J, Schubert N, Alam D, Samadi SD, Smith PL. The hypotonic upper airway in obstructive sleep apnea: role of structures and neuromuscular activity. Am J Respir Crit Care Med. 1998;157:1051–1057. doi: 10.1164/ajrccm.157.4.9706067. [DOI] [PubMed] [Google Scholar]
  40. Schwartz AR, Thut DC, Brower RG, Gauda EB, Roach D, Permutt S, Smith PL. Modulation of maximal inspiratory airflow by neuromuscular activity: effect of CO2. J Appl Physiol. 1993;74:1597–1605. doi: 10.1152/jappl.1993.74.4.1597. [DOI] [PubMed] [Google Scholar]
  41. Scott CH, Sutton MSJ, Gusani N, Fayad Z, Kraitchman D, Keane MG, Axel L, Ferrai VA. Effect of dobutamine on regional left ventricular function measured by tagged magnetic resonance imaging in normal subjects. Am J Cardiol. 1999;83:412–417. doi: 10.1016/s0002-9149(98)00879-0. [DOI] [PubMed] [Google Scholar]
  42. Shelton KE, Woodson H, Gay SB, Suratt PM. Pharyngeal fat in obstructive sleep apnea. Am Rev Respir Dis. 1993;148:462–466. doi: 10.1164/ajrccm/148.2.462. [DOI] [PubMed] [Google Scholar]
  43. Shepard JW, Jr, Gefter WB, Guilleminault C, Hoffman EA, Hoffstein V, Hudgel DW, Suratt PM, White DP. Evaluation of the upper airway in patients with obstructive sleep apnea. Sleep. 1991;14:361–371. doi: 10.1093/sleep/14.4.361. [DOI] [PubMed] [Google Scholar]
  44. Shepard JW, Jr, Thawley SE. Localization of upper airway collapse during sleep in patients with obstructive sleep apnea. Am Rev Respir Dis. 1990;141:1350–1355. doi: 10.1164/ajrccm/141.5_Pt_1.1350. [DOI] [PubMed] [Google Scholar]
  45. Singer JD. Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. J Educ Behav Stat. 1998;24:323–355. [Google Scholar]
  46. Stone M, Davis EP, Douglas AS, NessAiver M, Gullapalli R, Levine WS, Lundberg A. Modeling the motion of the internal tongue from tagged cine-MRI images. J Acoust Soc Am. 2001;109:2974–2982. doi: 10.1121/1.1344163. [DOI] [PubMed] [Google Scholar]
  47. Strohl KP, Fouke JM. Dilating forces on the upper airway in anaesthetized dogs. J Appl Physiol. 1985;58:452–458. doi: 10.1152/jappl.1985.58.2.452. [DOI] [PubMed] [Google Scholar]
  48. Strohl KP, Wolin AD, van Lunteren E, Fouke JM. Assessment of muscle action on upper airway stability in anesthetized dogs. J Lab Clin Med. 1987;110:221–230. [PubMed] [Google Scholar]
  49. Suratt PM, Dee P, Atkinson RL, Armstrong P, Wilhoit SC. Fluoroscopic and computed tomographic features of the pharyngeal airway in obstructive sleep apnea. Am Rev Respir Dis. 1983;127:487–492. doi: 10.1164/arrd.1983.127.4.487. [DOI] [PubMed] [Google Scholar]
  50. Wheatley JR, Kelly WT, Tully A, Engel LA. Pressure-diameter relationships of the upper airway in awake supine subjects. J Appl Physiol. 1991;70:2242–2251. doi: 10.1152/jappl.1991.70.5.2242. [DOI] [PubMed] [Google Scholar]
  51. Winer BJ, Brown DR, Michels KM. Statistical Principles in Experimental Design. New York: McGraw-Hill; 1991. [Google Scholar]
  52. Yeon SB, Reichek N, Tallant BA, Lima JAC, Calhoun LP, Clark NR, Hoffman EA, Ho KKL, Axel L. Validation of in vivo myocardial strain measurement by magnetic resonance tagging with sonomicrometry. J Am Coll Cardiol. 2001;38:555–561. doi: 10.1016/s0735-1097(01)01397-3. [DOI] [PubMed] [Google Scholar]
  53. Young T, Palta M, Dempsey J, Skatrud J, Weber S, Badr S. The occurrence of sleep-disordered breathing among middle-aged adults. N Engl J Med. 1993;328:1230–1235. doi: 10.1056/NEJM199304293281704. [DOI] [PubMed] [Google Scholar]
  54. Yuan Q, Axel L, Hernandez EH, Dougherty L, Pila JJ, Scott CH, Ferrari VA, Blom AS. Cardiac-respiratory gating method for magnetic resonance imaging of the heart. Magn Reson Med. 2000;43:314–318. doi: 10.1002/(sici)1522-2594(200002)43:2<314::aid-mrm21>3.0.co;2-9. [DOI] [PubMed] [Google Scholar]

Articles from The Journal of Physiology are provided here courtesy of The Physiological Society

RESOURCES