Coordinate Mapping of Hyolaryngeal Mechanics in Swallowing

Summary

Coordinate mapping is a method of documenting salient features of hyolaryngeal biomechanics in the pharyngeal phase of swallowing. This methodology uses image analysis software to record coordinates of anatomical landmarks. These coordinates are imported into an excel macro and translated into kinematic variables of interest useful in dysphagia research.

Abstract

Characterizing hyolaryngeal movement is important to dysphagia research. Prior methods require multiple measurements to obtain one kinematic measurement whereas coordinate mapping of hyolaryngeal mechanics using Modified Barium Swallow (MBS) uses one set of coordinates to calculate multiple variables of interest. For demonstration purposes, ten kinematic measurements were generated from one set of coordinates to determine differences in swallowing two different bolus types. Calculations of hyoid excursion against the vertebrae and mandible are correlated to determine the importance of axes of reference.

To demonstrate coordinate mapping methodology, 40 MBS studies were randomly selected from a dataset of healthy normal subjects with no known swallowing impairment. A 5 ml thin-liquid bolus and a 5 ml pudding swallows were measured from each subject. Nine coordinates, mapping the cranial base, mandible, vertebrae and elements of the hyolaryngeal complex, were recorded at the frames of minimum and maximum hyolaryngeal excursion. Coordinates were mathematically converted into ten variables of hyolaryngeal mechanics.

Inter-rater reliability was evaluated by Intraclass correlation coefficients (ICC). Two-tailed t-tests were used to evaluate differences in kinematics by bolus viscosity. Hyoid excursion measurements against different axes of reference were correlated. Inter-rater reliability among six raters for the 18 coordinates ranged from ICC = 0.90 - 0.97. A slate of ten kinematic measurements was compared by subject between the six raters. One outlier was rejected, and the mean of the remaining reliability scores was ICC = 0.91, 0.84 - 0.96, 95% CI. Two-tailed t-tests with Bonferroni corrections comparing ten kinematic variables (5 ml thin-liquid vs. 5 ml pudding swallows) showed statistically significant differences in hyoid excursion, superior laryngeal movement, and pharyngeal shortening (p < 0.005). Pearson correlations of hyoid excursion measurements from two different axes of reference were: r = 0.62, r² = 0.38, (thin-liquid); r = 0.52, r² = 0.27, (pudding).

Obtaining landmark coordinates is a reliable method to generate multiple kinematic variables from video fluoroscopic images useful in dysphagia research.

Introduction

The pharyngeal phase of swallowing is a complex process involving over twenty muscles and multiple skeletal elements to transfer a bolus from the oral cavity into the esophagus while protecting the airway. Preceding pharyngeal constriction, elements of the hyolaryngeal complex (hyoid bone, larynx, and associated structures including the upper esophageal sphincter) are displaced to convert a respiratory conduit into a digestive tract. The larynx is re-located anteriorly away from the trajectory of an oncoming bolus, and the upper esophageal sphincter is stretched open. Kinematic measurements taken from video fluoroscopic swallowing studies (also known as an MBS or Modified Barium Swallow Studies) are the primary research methodology for quantifying the multiple movements of the hyolaryngeal complex ¹.

While quantitative video fluoroscopic measurements are useful for measuring swallowing function, different axes of reference and scalars of measurement result in findings that are incompatible among the various methods of kinematic measurements ². The movement of the patient and fluoroscope under manual clinician control also confounds the accuracy of measuring this complex physiological process. More importantly, kinematic measurements do not necessarily reflect structure-to-function relationships important for evaluating disordered swallowing. Kinematics of the hyoid in particular have been designed to track movement in an anterior or superior direction in reference to an anatomical plane aligned with the vertebrae. However, this configuration does not represent the line of action of muscles that suspend the hyoid.

A two-sling mechanism of hyolaryngeal elevation in the pharyngeal phase of swallowing has been identified (Figure 1) ^3,4. The suprahyoid muscles comprise the anterior muscular sling, and the long pharyngeal muscles comprise the posterior muscular sling. These muscles elevate various elements of the hyolaryngeal complex including the hyoid, larynx, and structures that form the upper esophageal sphincter.

Coordinate mapping of hyolaryngeal mechanics utilizes nine easily identifiable anatomical landmarks to map three skeletal levers and features of the hyolaryngeal complex representing attachment points of the anterior and posterior muscular slings (Figure 2). During swallowing, each skeletal lever and feature of the hyolaryngeal complex is in motion. By collecting coordinates, the system can be captured in any time frame. Trigonometric conversion of coordinates can be used to generate multiple kinematic measurements of hyolaryngeal movement during swallowing. Variables can be calculated for comparison with findings reported in the literature, or used to generate new measurements representing structure-to-function relationships of interest.

The primary aim of this paper is to demonstrate a method of generating multiple kinematic measurements calculated from a single set of anatomical landmark coordinates collected from Modified Barium Swallow (MBS) studies. We document the reliability of this method by using Intraclass correlation coefficients to determine the inter-rater reliability of 6 different investigators including one expert, three raters with experience, and two novices. From the kinematic results, differences in swallowing mechanics by bolus consistency are evaluated. Finally, the question proposed by Molefenter and Steele concerning the importance of the axis of reference used in measuring hyoid movement is addressed. To approach this question we compare measurements of hyoid excursion in reference to the vertebrae and in reference to the mandible, calculated from the same set of coordinates for both bolus types. If these two methods of measuring hyoid movement represent the same structure to function relationship, then the results should be strongly correlated.

For this study, 40 lateral view MBS studies were randomly selected from a collection of 139 normal studies under a research protocol approved by the Georgia Regents University Institutional Review Board and in research collaboration with the Evelyn Trammell Institute for Voice and Swallowing at the Medical University of South Carolina. To demonstrate the utility of this method, ten variables characterizing hyolaryngeal kinematics were calculated from the same set of coordinate data (Table 1). Seven of these calculated measurements have previously been used in the literature including: anterior and superior hyoid distance measurements in reference to the vertebrae ⁵; anterior and superior hyoid displacement as a ratio of C2-4 length, also in reference to the vertebrae ⁶; superior movement of the larynx in reference to the vertebrae ⁷; hyolaryngeal approximation ¹; and maximum hyoid excursion in reference to the vertebrae ¹. In addition, three novel measurements were calculated: pharyngeal shortening approximating the attachments of the palatopharyngeus muscle, laryngeal elevation following a line of action representing the stylopharyngeus, and hyoid excursion approximating the attachments of the suprahyoid muscles ^4,8.

An expert head and neck anatomist (WP), three investigators with limited experience taking measurements (CJ, SR, TT), and two novice investigators (RS, JT) obtained coordinate mapping data using the protocol described below. The expert (WP) trained the three raters with experience, and these in turn trained the two novice raters. Inter-rater reliability of coordinate data and results calculated from coordinates by subject was determined by Intraclass correlation coefficients ⁹. Two tailed t-tests were performed on each variable to determine statistically significant differences in bolus types. A Pearson correlation coefficient and a coefficient of determination were used to evaluate agreement between results of hyoid excursion calculated with the vertebrae as an axis of reference versus hyoid excursion with the mandible as an axis of reference for the 5 ml thin-liquid swallow and 5 ml pudding swallow.

Protocol

1. Configuring a computer

1. For Macintosh, download the following open source or freeware software: ImageJ, MacX Video Converter Free Edition (Mac), and QuickTime (see table of material/equipment).
2. For a PC, download the following open source or freeware software: ImageJ, MPEG Streamclip (PC), and QuickTime (see table of material/equipment).

2. Preparing video clips

1. File conversion. Convert raw video files into .mov for data collection in ImageJ. Note: usable videos should include the mandible, hard palate, C1-C4 vertebrae, larynx and pharynx in the field of view. ImageJ processing tools can improve image quality issues.
   1. With a PC, use MPEG Streamclip, load the raw .avi video file via File >> Open Files and selecting the desired clip. Select File >> Export to QuickTime.
   2. With a Mac, use MacX Video Converter, click "Add File" and choose video(s) for conversion. Multiple videos can be converted at once using this application.
   3. In the Output Settings section, click the browse button and choose the desired location to save the converted file.
   4. Click on the "to MOV" tab. Make sure the Output Format under the "Video Settings" section is MOV. Click Start.
2. Editing clip length. Memory is limited with ImageJ. It is recommended that a file of each swallow be made.
   1. Open MOV. Video file created in step 2.1 using QuickTime (PC or Mac).
   2. Identify 5 ml thin-liquid and pudding swallows by audio queues or by swallow sequence outlined by the MBS protocol used during data collection.
   3. Select Edit >> Trim and adjust the trimming bar so that the entire 5 ml Thin Fluid swallow is visualized. Click trim.
   4. Select File >> Export and create a file name that will be used to link subject data (gender, age, etiology, bolus type) to coordinate mapping results.
   5. Repeat for the pudding swallow (or for any other bolus of interest).

3. De-identifying images

1. If a file contains any personal health information (PHI) and needs to be de-identified, upload the file using Image J (See 4.1 - 4.2).
2. Use the rectangle tool to frame the swallowing study to exclude PHI. Select Image >> Crop. Then Select File >> Save As >> QuickTime Movie.
3. Configure the dialogue box as follows: Compression >> Sorenson 3; Quality >> Maximum, enter the appropriate frame rate (usually 30 fps). 

4. Preparing to measure

1. Open ImageJ. Click on the ">>" icon on the toolbar. Select >> Arrow Labeling Tools.
2. Uploading images. Click on QuickTime icon. Select "Open movie as a stack" from the drop down menu and locate edited QuickTime clip. Image J will not open the entire swallow study due to memory limits (see step 2.2). Memory can be minimally expanded by selecting Edit >> Options >> Memory & Threads.
3. Processing images to improve image quality. Select Process >> Math >> Add. Check the preview box and adjust numbers to desired image quality. Reply yes to process the entire stack of images.
4. Set measurements. Select Analyze >> Set Measurements from the ImageJ menu. At the dialogue box mark "Stack Position" and "Invert Y coordinates". Unmark everything else.
5. Select the multipoint tool from the toolbar. Click on anatomical landmarks in sequence (see 5.0)
6. Using the multipoint tool.
   1. Take a measurement of all points by selecting Analyze >> Measure from the menu or keyboard command+M (control+M for PCs).
   2. Remove all points with command+A (control+A for PCs).
   3. Remove single points by hovering over a point, then hold the command-option keys and click on the point to be removed.
   4. Move singular points by hovering over a point, clicking, dragging and dropping a point to a new location.
   5. Move all points together with arrow keys.

5. Mapping Landmarks

1. Begin at first frame and advance to a clear frame in the pre-oral phase. Observe position of the bolus on the anterior, superior margin of the tongue prior to initiation of oral transport of the swallow. Use this frame to set the first nine coordinates.
2. Use the ImageJ multi-point tool to map the first nine coordinates at the following anatomical landmarks (see Figures 3a and 3b)
3. Record first nine coordinates using command+M keys (control+M for PCs).
4. Advance frames until hyoid bone has reached maximum position in anterior and superior directions. Confirm maximum by advancing frames to insure descent of the hyoid begins on the following frame.
5. Relocate points 1 - 5 to their new positions. These new positions will be recorded as coordinates 10 - 14.
6. Relocate Point 9, which in turn becomes coordinate 18. Note: frames will likely vary for the next two steps.
7. Locate frame depicting maximum laryngeal elevation. Adjust points 7 and 8, which will serve as coordinates 16 and 17.
8. Find frame(s) representing the maximum excursion of UES, point 6 (coordinate 15). From maximum hyoid frame, locate frame where the bolus is impeded by the UES in the hypopharynx. Adjust coordinate point for UES from minimum frame to represent UES maximum coordinate 18.
9. Record second nine coordinates using command+M keys (control+M for PCs).
10. For coordinates 19 and 20 mark the edges of the scalar (a penny or 1.9 cm ring) at the axis representing the longest diameter of the radiopaque marker.
11. Record scalar coordinates using command+M keys (control+M for PCs).

6. Transform coordinate data into kinematic measurements of interest using a macro enabled Excel file (the instructions for this macro are included on the file)

Note: Trigonometric calculations imbedded in the macros calculate kinematic measurements (Figure 4).

1. Download "CoordinateMapping.xlsm" from the JoVE article page.
2. Follow instructions on the spreadsheet to initialize the file. The initialization macro will create three sheets including: results, data, and input sheet. Note: Macintosh users must enable developer tools in Excel to run macros.
3. Copy coordinates from the ImageJ results window and paste into the designated cell in the "input sheet". Run the "datacaptureline" macro.
4. Results will appear on the "results" sheet. Lines of coordinate data will appear on the "data" sheet.

Results

Intraclass correlation coefficients (ICC) of coordinates collected by six investigators who independently analyzed 80 video fluoroscopic files (two bolus trial from 40 subjects) ranged from ICC = 0.90 - 0.97. A breakdown of ICC's of coordinates by group is as follows: coordinates #1 - 5 (skeletal elements at minimum hyolaryngeal excursion) mean = 0.93, 0.91 - 0.95, 95% CI; coordinates #6 - 9 (hyolaryngeal complex at minimum hyolaryngeal excursion) mean = 0.94, 092 - 0.96, 95% CI; coordinates #10 - 14 (skeletal elements a...

Discussion

This study demonstrates the usefulness of a method using coordinate data of anatomical landmarks to calculate multiple kinematic measurements of hyolaryngeal movement in swallowing. Inter-rater reliability of six raters, including two novice raters, for coordinates and calculated variables was strong (ICC > 0.90). Representative results from a random sample of healthy non-dysphagic adults showed differences in several kinematic variables in response to two bolus types. We also found that using different axes of refer...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
ImageJ	NIH	http://rsbweb.nih.gov/ij/download.html	For Macintosh
MacX Video Converter Free Edition (Mac)	Digiarty	http://www.macxdvd.com/mac-video-converter-free/	For Macintosh
QuickTime	Apple	http://support.apple.com/downloads/#QuickTime	For Macintosh
ImageJ	NIH	http://rsbweb.nih.gov/ij/download.html	For a PC
MPEG Streamclip (PC)	Squared 5	http://www.squared5.com	For a PC
QuickTime	Apple	http://support.apple.com/downloads/#QuickTime	For a PC