You may notice some images loading slow across the Open Collections website. Thank you for your patience as we rebuild the cache to make images load faster.

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The effects of muscle aging on hyoid motion during swallowing : a study using a 3D biomechanical model Tsou, Ling 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


24-ubc_2012_fall_tsou_ling.pdf [ 17.42MB ]
JSON: 24-1.0072854.json
JSON-LD: 24-1.0072854-ld.json
RDF/XML (Pretty): 24-1.0072854-rdf.xml
RDF/JSON: 24-1.0072854-rdf.json
Turtle: 24-1.0072854-turtle.txt
N-Triples: 24-1.0072854-rdf-ntriples.txt
Original Record: 24-1.0072854-source.json
Full Text

Full Text

The effects of muscle aging on hyoid motion during swallowing: A study using a 3D biomechanical model by Ling Tsou  B.Eng., McMaster University, 2009  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in The Faculty of Graduate Studies (Electrical and Computer Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) June 2012 c Ling Tsou 2012  Abstract The ability to swallow is crucial in maintaining adequate nutrition. However, there is a high prevalence of dysphagia among the elderly and a high associated mortality rate. To study the various causes of the associated physiological changes, one must first understanding the biomechanics of normal swallowing. However, functional studies of the anatomically complex head and neck region can prove to be difficult due to both technical and ethical reasons. To overcome the limitations of clinical studies, this thesis proposes the use of a 3D computer model for performing dynamic simulations. A stateof-the-art model of the hyolaryngeal complex was created for simulating swallowing-related motor tasks with a special focus on hyoid excursion since reduced hyoid motion is a major indicator of oropharyngeal dysphagia. The model was constructed using anatomical data for a male cadaver from the Visible Human Project c and an open-source dynamic simulation platform, ArtiSynth. Hyoid motion data obtained from videofluoroscopy of subjects performing normal swallowing was applied to the model for inversely simulating the potential muscle activities of the extrinsic laryngeal muscles during hyoid excursion. Within a specific range, the model demonstrated the ability to reproduce realistic hyoid motion for swallowing. Selective usage of suprahyoid muscles was also examined and was found to be possible in achieving adequate hyoid excursion for successful swallows. Finally, this study investigated the relationship between muscle weakening and hyoid range of motion using the hyolaryngeal model. Loss of muscle strength is characteristic of the aging process. Simulation of the maximum hyoid displacement under various muscle conditions confirmed ii  Abstract a nonlinear reduction in the hyoid motion range under a linear decline in muscle strength. With an assumed rate of muscle weakening, the proportion of hyoid range reduction was estimated for a person at various ages. The results suggest that severe muscle weakening might be required to reduce hyoid excursion sufficiently to impair swallowing to a significant degree.  iii  Preface This thesis was part of the Oral, Pharyngeal, and Laryngeal Complex (OPAL) project and parts of Chapter 4 were published in L. Tsou and S. Fels. Biomechanical modeling of the external laryngeal structures. In International Symposium on Speech Sciences (ISSP’2011), pages 133–138, 2011 [130]. The magnetic resonance image included in Chaper 1 Figure 1.2 was obtained under ethical approval from UBC Clinical Research Ethics Board, Certificate number H11-00375 and H07-01824. All fluoroscopic images included in this thesis were provided by Christy Ludlow of the National Institute of Neurological Disorders and Stroke and were published in I. Stavness, C. L. Ludlow, B. Chung, and S. Fels. Hyolaryngeal biomechanics modeling with intramuscular stimulation data. In 1st International Workshop on Dynamic Modeling of the Oral, Pharyngeal and Laryngeal Complex for Biomedical Applications (OPAL-2009), pages 73–78, Vancouver, BC, Canada, 2009 [122].  iv  Table of Contents Abstract  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ii  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  iv  Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . .  v  Preface  List of Tables  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii  List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Glossary  ix  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii  Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1  Motivation  . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1.2  Problem Statement and Proposed Solution  1.3  Contributions  1.4  1 4  . . . . . . . . . .  5  . . . . . . . . . . . . . . . . . . . . . . . . . .  6  Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  7  2 Related Research . . . . . . . . . . . . . . . . . . . . . . . . . .  8  2.1  Hyoid Motion and Dysphagia . . . . . . . . . . . . . . . . . .  8  2.2  Hyoid Excursion . . . . . . . . . . . . . . . . . . . . . . . . .  10  2.3  Swallowing and Aging . . . . . . . . . . . . . . . . . . . . . .  13  2.4  Muscle Aging . . . . . . . . . . . . . . . . . . . . . . . . . . .  14  2.5  Existing Computer Models  16  . . . . . . . . . . . . . . . . . . .  v  Table of Contents 3 Model of the Hyolaryngeal Complex . . . . . . . . . . . . . .  18  3.1  Bones and Cartilages  . . . . . . . . . . . . . . . . . . . . . .  20  3.2  Muscles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  22  3.3  Ligaments and membranes  . . . . . . . . . . . . . . . . . . .  26  3.4  Joint  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  29  4 Implementation of the Model  . . . . . . . . . . . . . . . . . .  30  4.1  Data Source and Segmentation . . . . . . . . . . . . . . . . .  31  4.2  Making Symmetry . . . . . . . . . . . . . . . . . . . . . . . .  33  4.3  Volumetric Mesh Fitting  . . . . . . . . . . . . . . . . . . . .  39  . . . . . . . . . . . . . . . . . . . . . . . .  41  4.3.1  Templates  4.3.2  Fitting  . . . . . . . . . . . . . . . . . . . . . . . . . .  42  4.3.3  Comparison Between the Two Templates . . . . . . .  49  4.4  Muscle Attachment  . . . . . . . . . . . . . . . . . . . . . . .  4.5  Muscle Fiber Specification  4.6  Coupling  52  . . . . . . . . . . . . . . . . . . .  53  . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  55  4.6.1  Connective Tissues as FEM Meshes . . . . . . . . . .  57  4.6.2  Active and Passive Springs and Spring Networks . . .  57  4.6.3  Constraints . . . . . . . . . . . . . . . . . . . . . . . .  58  4.7  Mechanical Properties Specification  . . . . . . . . . . . . . .  58  4.8  Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  58  5 Evaluation I: Muscle Functions and Swallowing Simulation 60 5.1  Individual Muscle Function Simulation  5.2  Inverse Simulation of Swallowing Motion  . . . . . . . . . . . .  60  . . . . . . . . . . .  65  5.2.1  Data Source  . . . . . . . . . . . . . . . . . . . . . . .  66  5.2.2  Inverse Controller . . . . . . . . . . . . . . . . . . . .  67  5.2.3  Methodology . . . . . . . . . . . . . . . . . . . . . . .  68  5.2.4  Results . . . . . . . . . . . . . . . . . . . . . . . . . .  70  5.2.5  Discussion  78  . . . . . . . . . . . . . . . . . . . . . . . .  6 Evaluation II: Hyoid Motion and Muscle Aging  . . . . . .  81  6.1  Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . .  81  6.2  Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . .  82 vi  Table of Contents 6.3  Results  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  6.4  Discussion  84  . . . . . . . . . . . . . . . . . . . . . . . . . . . .  84  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  88  7.1  Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  88  7.2  Future Directions  . . . . . . . . . . . . . . . . . . . . . . . .  89  Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  92  7 Conclusion  Appendices A Anatomy and Swallowing Physiology  . . . . . . . . . . . . . 107  A.1 Anatomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 A.1.1 Airway  . . . . . . . . . . . . . . . . . . . . . . . . . . 107  A.1.2 Bones, Cartilages and Connective Tissues . . . . . . . 107 A.1.3 Muscles . . . . . . . . . . . . . . . . . . . . . . . . . . 112 A.2 Swallowing Physiology  . . . . . . . . . . . . . . . . . . . . . 115  A.2.1 Oral Phase . . . . . . . . . . . . . . . . . . . . . . . . 115 A.2.2 Pharyngeal Phase . . . . . . . . . . . . . . . . . . . . 115 A.2.3 Esophageal Phase . . . . . . . . . . . . . . . . . . . . 115 B Mechanical Properties . . . . . . . . . . . . . . . . . . . . . . . 116 B.1 Rigid Bodies . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 B.2 Finite Element Models  . . . . . . . . . . . . . . . . . . . . . 116  B.2.1 Mooney Rivlin Material . . . . . . . . . . . . . . . . . 117 B.2.2 Linear Material B.3 Muscle Model  . . . . . . . . . . . . . . . . . . . . . 117  . . . . . . . . . . . . . . . . . . . . . . . . . . 117  B.4 Spring Network C Additional Material  . . . . . . . . . . . . . . . . . . . . . . . . . 119 . . . . . . . . . . . . . . . . . . . . . . . . 120  vii  List of Tables 3.1  Rigid structures in the model . . . . . . . . . . . . . . . . . .  23  3.2  Muscles in the model . . . . . . . . . . . . . . . . . . . . . . .  24  3.3  Connective tissues in the model . . . . . . . . . . . . . . . . .  28  5.1  Landmark definitions . . . . . . . . . . . . . . . . . . . . . . .  67  5.2  Evaluation I inverse simulation trials: experiment 1 . . . . . .  69  5.3  Evaluation I inverse simulation trials: experiment 2 . . . . . .  70  6.1  Evaluation II simulation trials . . . . . . . . . . . . . . . . . .  83  A.1 Extrinsic laryngeal muscles . . . . . . . . . . . . . . . . . . . 114 B.1 Rigid body properties . . . . . . . . . . . . . . . . . . . . . . 116 B.2 Linear connective tissue properties . . . . . . . . . . . . . . . 118 B.3 Muscle properties . . . . . . . . . . . . . . . . . . . . . . . . . 118 B.4 Spring network properties . . . . . . . . . . . . . . . . . . . . 119 C.1 Hyoid displacement reported by past studies . . . . . . . . . . 121  viii  List of Figures 1.1  Dysphagia classification . . . . . . . . . . . . . . . . . . . . .  3  1.2  Examples of swallowing imaging techniques . . . . . . . . . .  4  2.1  The swallowing process  . . . . . . . . . . . . . . . . . . . . .  9  2.2  Hyoid excursion . . . . . . . . . . . . . . . . . . . . . . . . . .  11  2.3  Examples of posture and intersubject variability . . . . . . .  12  2.4  Examples of existing upper airway models . . . . . . . . . . .  17  3.1  Model of the hyolaryngeal complex . . . . . . . . . . . . . . .  19  3.2  Rigid body model . . . . . . . . . . . . . . . . . . . . . . . . .  22  3.3  FEM and spring muscle model . . . . . . . . . . . . . . . . .  24  3.4  Pharyngeal constrictors modeled as spring networks . . . . .  26  3.5  Connective tissues in the model . . . . . . . . . . . . . . . . .  27  3.6  Omohyoid intermediate tendon sheath modeled . . . . . . . .  27  3.7  Cricothyroid articulation . . . . . . . . . . . . . . . . . . . . .  29  4.1  Process flow of biomechanical model generation . . . . . . . .  30  4.2  c  Surface geometries segmented from the Visible Human Project  datasets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  32  4.3  Labeling of some cryosection images . . . . . . . . . . . . . .  32  4.4  Illustration of the symmetry making process . . . . . . . . . .  35  4.5  Example of making geometry symmetric . . . . . . . . . . . .  37  4.6  The process of making the model symmetric . . . . . . . . . .  38  4.7  Examples of volumetric mesh . . . . . . . . . . . . . . . . . .  39  4.8  Two types of mesh template . . . . . . . . . . . . . . . . . . .  42  ix  List of Figures 4.9  Graphical illustration of the template fitting technique: first four steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  44  4.10 Illustration of rectangular grid fitting . . . . . . . . . . . . . .  45  4.11 Illustration of finding the centroid of a 2D shape . . . . . . .  46  4.12 Illustration of circular grid fitting . . . . . . . . . . . . . . . .  48  4.13 Rectangular and circular grid fitting example . . . . . . . . .  49  4.14 Template fitting of an idealized muscle . . . . . . . . . . . . .  50  4.15 Hexahedral mesh fitting on the digastric anterior muscle using the beam template . . . . . . . . . . . . . . . . . . . . . . . .  51  4.16 Example of inverted elements: digastric muscle . . . . . . . .  52  4.17 Attachment of the sternohyoid muscle . . . . . . . . . . . . .  53  4.18 Muscle attachments . . . . . . . . . . . . . . . . . . . . . . .  54  4.19 Fiber specification of the thyrohyoid muscle . . . . . . . . . .  55  4.20 Two types of muscle models . . . . . . . . . . . . . . . . . . .  56  4.21 Ligaments of the larynx . . . . . . . . . . . . . . . . . . . . .  57  5.1  Muscle activation waveform for evaluation I . . . . . . . . . .  61  5.2  Mid-sagittal view of the larynx at rest . . . . . . . . . . . . .  62  5.3  Actions of the suprahyoid muscles . . . . . . . . . . . . . . .  63  5.4  Actions of the infrahyoid muscles . . . . . . . . . . . . . . . .  64  5.5  Actions of the thyrohyoid muscles . . . . . . . . . . . . . . . .  64  5.6  Group actions of the extrinsic laryngeal muscles . . . . . . . .  65  5.7  Landmarks in the videofluoroscopy and model . . . . . . . . .  67  5.8  Evaluation I inverse simulation experiment 1 results: hyoid trajectories . . . . . . . . . . . . . . . . . . . . . . . . . . . .  5.9  72  Evaluation I inverse simulation experiment 1 results: muscle activations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  73  5.10 Evaluation I inverse simulation experiment 1 results: error versus distance . . . . . . . . . . . . . . . . . . . . . . . . . .  74  5.11 Model in motion . . . . . . . . . . . . . . . . . . . . . . . . .  75  5.12 Evaluation I inverse simulation experiment 2 results: hyoid trajectories . . . . . . . . . . . . . . . . . . . . . . . . . . . .  76  x  List of Figures 5.13 Evaluation I inverse simulation experiment 2 results: muscle activations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  77  5.14 Morphological difference between individuals. . . . . . . . . .  79  6.1  Muscle activation waveform for evaluation II . . . . . . . . .  83  6.2  Evaluation II results: maximum hyoid displacement vs. mus-  6.3  cle strength . . . . . . . . . . . . . . . . . . . . . . . . . . . .  85  Evaluation II results: maximum hyoid displacement vs. age .  86  A.1 The respiratory system . . . . . . . . . . . . . . . . . . . . . . 108 A.2 Upper airway anatomy . . . . . . . . . . . . . . . . . . . . . . 108 A.3 Cartilages of the larynx . . . . . . . . . . . . . . . . . . . . . 110 A.4 Side view of the skull . . . . . . . . . . . . . . . . . . . . . . . 110 A.5 Bones of the shoulder . . . . . . . . . . . . . . . . . . . . . . 111 A.6 Extrinsic laryngeal muscles . . . . . . . . . . . . . . . . . . . 112 A.7 Pharyngeal constrictor muscles . . . . . . . . . . . . . . . . . 113  xi  Glossary C2 Second Cervical Vertebrae C4 Fourth Cervical Vertebrae CASLPA Canadian Association of Speech-Language Pathologists and Audiologists CT Computed Tomography EMG Electromyography FEM Finite-Element Method HA Anterior Hyoid MRI Magnetic Resonance Imaging PCSA Physiological Cross-Sectional Area VHP Visible Human Project VF Videofluoroscopy UES Upper Esophageal Sphincter  xii  Acknowledgements I would like to thank my supervisor Dr. Sidney Fels for his guidance and support. Thanks to Ian Stavness, John Loyd, and all those at the Human Communication Technologies Laboratory for their continued assistance. Finally, thanks to my family and friends for their love and support.  xiii  Chapter 1  Introduction The ability to swallow, that is, a process to transport food from the mouth to the stomach, is crucial for a person to maintaining adequate nutrition. The symptom where a patient experiences difficulty in swallowing is known as dysphagia in the medical world. Dysphagia at different levels of severity can have various consequences. Minor signs include coughing, choking, and globus sensation, but more serious consequences include malnutrition, dehydration, and reduced quality of life [31]. Aspiration can be considered one of the most dangerous consequences of dysphagia [77]. Failure to prevent objects from entering the trachea could lead to pneumonia, which can be fatal if not treated. Dysphagia can result from numerous medical conditions, which can be neurological (such as Parkinson’s disease), myopathic (such as myotonic dystrophy), structural (such as oropharyngeal tumors), and even iatrogenic (such as chemotherapy and radiation) [18, 38, 73]. Due to the wide range of causes, studies on dysphagia most often involve not only otolaryngologists but also dentists, neurologists, speech pathologists, radiologists, and sometimes medical researchers from other domains. Among the elderly, dysphagia has an especially high prevalence. It’s estimated that about one in 10 people over 50 have some type of swallowing disorder, which experts call dysphagia. —CASLPA Fact Sheet. One study reported a 12-month mortality of 45% in nursing home patients with dysphagia and aspiration [20]. The importance of understanding dysphagia has lately received significant attention from both the academic and clinical communities. However, diagnosis and treatment of dysphagia in the 1  Chapter 1. Introduction elderly can prove to be challenging. Diagnosis of dysphagia is often difficult in the elderly due to a lowered ability to communicate. Complications with a systemic disease can also hamper treatment and research. These factors together constitute a major obstacle to research on the association between dysphagia and aging. Given that dysphagia often appears as a symptom of an underlying disease, it is not surprising that there is a high prevalence among the elderly. One may suspect the increased susceptibility to diseases in general is the main reason for this prevalence. While this might be a possible explanation, it may not be the sole reason. Besides making individuals more prone to disease, aging is also associated with changes in muscle physiology. Loss of muscle mass is one of the major indicators of late adulthood. Muscle aging alone is a topic that has garnered considerable interest and has been studied intensively. Swallowing is a complex process that consists of a series of voluntary and involuntary actions. A successful swallow requires precise and coordinated motions of numerous muscles. Changes in muscular performance will likely lead to changes in swallowing behavior. However, the direct relationship between muscle aging and dysphagia has only been speculated due to various difficulties, both technical and ethical. Dysphagia can be categorized into two types depending on the site of dysfunction: oropharyngeal and esophageal (see Figure 1.1). The former, which is the focus of this thesis, is commonly caused by functional aberrations, whereas the latter is often the result of mucosal or structural aberrations [18]. Therefore, one must first understand the physiology of normal oropharyngeal swallowing before proceeding to a study of oropharyngeal dysphagia. Swallowing itself is a highly complex and coordinated procedure that involves a series of very specialized movements of the oral, pharyngeal, laryngeal, and esophageal structures. Fine control of muscles in these areas is essential for a successful swallow. However, these complex anatomical structures are tightly coupled and deeply embedded in the head and neck region. These features make it difficult to closely examine their behaviors while in motion. Indeed, to study their behaviors during swallowing in vivo is extremely challenging, if not impossible, due to the size and location of the 2  Chapter 1. Introduction upper airway structures. Moreover, precise and specific movements of these tissues can rarely be acquired by modern dynamic imaging techniques such as magnetic resonance imaging (MRI) (Figure 1.2). Adding to the difficulty in communication and complications with other diseases, these obstacles are only more problematic when investigating oropharyngeal dysphagia in the elderly.  Figure 1.1: Dysphagia is classified into two types, oropharyngeal and CC esophageal, depending on the site of dysfunction. by Olek Remesz via Wikimedia Commons.  There are many anatomical and physiological factors that can contribute to oropharyngeal dysphagia. Among them, the range of hyoid motion during swallowing has been reported as one of the major indicators of oropharyngeal dysphagia [78]. Reduced and delayed laryngeal elevation are the most common causes of aspiration among dysphagia patients [73]. Statistically speaking, a reduced range of hyoid and laryngeal motion has been observed in both the healthy elderly [60, 139] and dysphagia patients [73, 90]. This thesis focuses on examining the changes in hyoid motion that result from muscle aging. While not directly studying its motion during the swallowing process, the range of hyoid motion can serve as a risk factor for oropha-  3  1.1. Motivation  (a) MRI  (b) Fluoroscopy  Figure 1.2: Examples of some dynamic imaging techniques. (b) Fluoroscopic image by [122].  ryngeal dysphagia. Although research on this topic is challenging for the many reasons outlined above, computational modeling can serve as an important tool for overcoming these difficulties. Here a 3D dynamic model of the hyolaryngeal complex was developed for simulating hyoid movements under various muscle conditions. The model facilitates investigation and prediction of the relationship between the range of hyoid motion and muscle weakening. Thus, further understanding into the effects of muscle aging on oropharyngeal dysphagia could be gained.  1.1  Motivation  Researchers have observed that across age groups, even subjects without any dysfunction demonstrate certain differences in the swallowing process [30, 41, 105]. There were differences in both the timing and range of motion throughout various phases of swallowing. It has been suggested that these differences might contribute to the risk of swallowing dysfunction in the elderly. However, the causes of these changes remains an open question. After all, the process of aging is complicated and cannot be easily understood  4  1.2. Problem Statement and Proposed Solution and quantified. A number of physiological changes occur throughout the process of aging. There are changes in bone density, tissue properties, muscle physiology, and the nervous system, just to name a few. Changes in muscle physiology are some of the factors that can directly impact the performance of a biomechanical system such as that involved in swallowing. As evidence to support this theory, dysphagia is frequently reported in patients with myopathic diseases such as inflammatory myopathy and muscular dystrophy [85, 137] where abnormal swallowing is caused by muscle weakening, and it is well established that muscular performance declines with advancing age [101]. However, the process of muscle aging varies considerably across individuals. Gender, diet, lifestyle, genetics, and medical conditions can all have an influence on this process. Moreover, there is also a high variability in the anatomical structures of the upper airway. Variations in size, shape, and relative position of the hyoid bone have been reported [81, 92, 125]. Naturally, the range of hyoid motion and its relationship with muscle performance are difficult to determine due to these variations. A long-term study on a large pool of subjects is likely required for research of this nature. Conversely, computer simulation provides a way to isolate individual variability and directly examine effects of muscle aging on hyoid motion.  1.2  Problem Statement and Proposed Solution  The goal of this thesis was to attempt to answer the question, What are the effects of muscle aging on the range of hyoid movement, specifically, the movements required during the pharyngeal phase of swallowing? The use of a 3D biomechanical model of the hyolaryngeal complex is proposed for simulating hyoid movement under various muscle conditions. Since muscle aging itself is a complicated process, the focus was on a specific aspect of muscle aging: muscle weakening. The muscle conditions were approximated to characterize loss of muscle strength across different age groups. A dynamic model utilizing the finite-element method (FEM) was de5  1.3. Contributions veloped from medical image data along with generic material properties gathered from the literature to model the intricate biomechanics of the hyolaryngeal complex, which consists of deformable muscle structures that are closely coupled with rigid bone structures. While the focus was to examine the maximum hyoid motion in relation to muscle strength decline, simulation of normal oropharyngeal swallow was also a part of this thesis for demonstrating that the model is capable of producing realistic hyolaryngeal motion.  1.3  Contributions  The contributions of this thesis revolve around the computer model developed for the purpose of studying swallowing, including the creation of the model, testing its functionality, and investigating its applicability in research on the process of swallowing and, specifically, age-related dysphasia. The primary contributions are listed below: 1. Creation of a 3D biomechanical model of the hyolaryngeal complex. A computer model was developed that consists of coupled rigid and soft tissue structures for simulating hyolaryngeal movement within the neck. 2. Identification of a workflow for computer model generation. A workflow is introduced for creating biomedical models of a nonspecific musculoskeletal system from medical images. 3. Inverse simulation of hyoid excursion during normal swallowing. The capability of the hyolaryngeal model to mimic realistic hyoid motion was demonstrated by performing inverse simulations using clinically measured motion data. The potential of selective muscle usage during swallowing was also examined. 4. Study of the effects of age-related muscle weakening on oropharyngeal swallowing. The relationship between loss of muscle strength and hyoid motion range was investigated. Based on an assumed rate 6  1.4. Outline of muscle weakening, the proportion of motion range reduction was estimated for a person at various ages. Secondary contributions include the development of a process for making surface geometries symmetric and implementation of a volumetric meshfitting technique.  1.4  Outline  The rest of this thesis is organized as the following. First, an overview of existing research in the relative fields of study is presented (Chapte 2). The hyolaryngeal model then is presented in Chapter 3, and the implementation procedure is covered in Chapter 4. The simulation of a normal oropharyngeal swallow is described in Chapter 5. Chapter 6 presents a study on the relationship between age-related muscle weakening and hyoid motion range. Finally, the thesis is concluded in Chapter 7 with a summary and some possible future directions. Background information on the anatomy and physiology related to the swallowing process is provided in Appendix A.  7  Chapter 2  Related Research 2.1  Hyoid Motion and Dysphagia  The extent of hyoid excursion is understood to have a close relationship with oropharyngeal dysphagia. Figure 2.1 demonstrates the process of a normal swallow, where the motion of the hyoid and larynx is highlighted in red. Perlman et al. [99] reported reduced hyoid displacement in both the anterior and vertical directions in dysphagia patients compared with normal subjects. Lundy et al. [73] further studied dysphagia patients who aspirated and found that reduced laryngeal elevation, which is largely contributed by hyoid motion due to its close coupling to the larynx, was the most common factor behind aspiration. Studies on dysphagia of specific etiology also showed similar reduction in hyoid motion in patients with diseases such as stroke, myopathy [90], and irradiated nasopharyngeal carcinoma [134]. A study by Eisbruch et al. [29] examined the effects on swallowing function of radiation therapy for head-and-neck cancer. They found a significantly increased occurrence rate of both aspiration and hyoid motion reduction in post-therapy patients, which suggests a significant degree of association between the two. Zu et al. [142] compared hyoid displacement in post-radiation and post-surgery head-and-neck cancer patients and argued that radiation therapy provides greater functional preservation potential compared with surgery since greater hyoid displacement was observed in post-radiation patients. Furthermore, MBSImp, a modified barium swallowing study tool introduced by Martin-Harris et al. [78], uses anterior hyoid motion and laryngeal elevation as parameters in its scoring system to quantify swallowing impairment. There are two main ways in which insufficient hyoid excursion can result 8  2.1. Hyoid Motion and Dysphagia  (a)  (b)  (c)  (d)  (e)  (f)  Figure 2.1: MRI of a normal liquid swallow showing different stages of the process. (a) and (b) oral phase where the bolus travels across the oral cavity; (c)–(e) pharyngeal phase where the larynx raises, epiglottis inverts to protect the airway, and the bolus moves from the pharynx into esophagus; and (f) return to rest. Note that the esophageal phase where the bolus is propelled down the esophagus is not shown here. The blue and red lines highlight the edges of the epiglottis and larynx respectively. CC by [131] via Wikimedia Commons.  9  2.2. Hyoid Excursion in an abnormal swallow. Epiglottic closure plays a crucial role in the airway protection mechanism for swallowing. While the underlying mechanism for the banding of the epiglottis is controversial [69, 132], the passive downward tilting of the epiglottis is widely recognized as being the result of hyoid excursion and extreme thyrohyoid approximation [2, 32, 35, 75]. Ardran and Kemp [2] stated that not only thyrohyoid approximation but also elevation of the hyoid bone and thyroid cartilage toward the mandible are essential for downfolding of the epiglottis. Another important function of hyoid excursion is to assist in upper esophageal sphincter (UES) opening. The musculature attached to the hyoid bone, together with hyoid excursion, pull the UES apart and initiate normal sphincter opening [17, 54]. In other words, inadequate hyoid displacement would likely cause failure in airway protection and bolus transport to the esophagus, and lead to laryngeal penetration or multiple swallows.  2.2  Hyoid Excursion  Hyoid excursion is generally characterized as a four-step motion: elevation, anterior motion, pause, and return (Figure 2.2). Numerous studies have attempted to understand hyoid motion for the purpose of swallowing. Table C.1 provides a list of past studies on hyoid excursion together with details on their subjects. Most of these studies were done using lateral projection videofluorography (VF) and barium-mixed bolus, which allows the hyoid bone to be easily identified. The maximum hyoid displacements in the anterior and superior directions were recorded and statistically analyzed across subjects. A relatively high variability in hyoid displacement during normal swallowing can be observed across studies. The anterior displacement ranges from 0.8 cm [70, 118] to over 1.6 cm [25, 60], and the superior displacement ranges from 0.5 cm [79, 118] to over 1.5 cm [25, 60, 71, 139]. A number of factors can contribute to these differences. The influence of calibration referents was examined by Sia et al. [114], who found no significant contribution to the variability. Molfenter and Steele [84] commented on potential 10  2.2. Hyoid Excursion  (a) Rest  (b) Upward  (c) Forward  (d) Return  Figure 2.2: Triangular pattern of hyoid excursion.  sources of variations as being statistical, methodological, stimulus-related, or participant-related. First of all, the axes that served as a reference for defining the vertical and horizontal measurements differed from one study to the next. The cervical spine was used to define the vertical axis in [23, 25], whereas [60, 90] specifically define the vertical axis by connecting the anterior corners of the second and forth cervical vertebrae. Ishida et al. [53] and Mays et al. [79] used the occlusal plane of the upper teeth as the horizontal axis. Other studies either directly used the image axes or did not specify how the axes were defined. Figure 2.3 provides an example showing how different the axes can be due to posture and intersubject variability. Another factor contributing to this inconsistency in maximum hyoid displacement is the bolus consumed. It is well established that hyoid movement increases with respect to bolus size [23, 25]. Gravity might also be a source of variation, depending on how the body was positioned, upright or supine. Lastly, 11  2.2. Hyoid Excursion Mays et al. [79] found that the anterior displacement of the hyoid was inversely correlated with the Frankfort-mandibular plane angle (FMA), which suggests that individual morphological differences can influence hyoid motion and swallowing physiology. Nonetheless, hyoid excursion during normal swallow can be quantified as having a range from around 0.8 to 1.8 cm for both maximum elevation and anterior movement.  (a)  (b)  Figure 2.3: Examples of fluoroscopic image from two different subjects demonstrating posture and morphological variability. Axes of measurements: axes in green were defined by the occlusal plane whereas axes in red were defined by the anterior corners of C2 and C4. Fluoroscopic images by [122].  The hyoid is attached to about 10 pairs of muscles, including the genioglossus, as shown in Section A.1.3. Among those muscles, the geniohyoid and mylohyoid have the most potential for moving the hyoid in the anterior and superior directions, respectively, according to a cadaver study done by Pearson et al.[94]. This partly matches the findings of Burnett et al. [14], who performed neuromuscular stimulation of selected muscles to examine their contribution to laryngeal motion. In their study, the paired mylohyoid together with the thyrohyoid accounted for about 50% of the laryngeal elevation and 80% of the velocity that was observed in a normal swallow. Unilateral stimulation of the mylohyoid, thyrohyoid, or geniohyoid muscle 12  2.3. Swallowing and Aging also accounted for around 30% of laryngeal elevation and 50% of the velocity. Electromyography (EMG) studies have shown large intersubject variability in both duration and sequence of suprahyoid muscle activation during swallowing [47, 120]. The muscles are also used selectively, where some subjects activate all while others activate different subsets of muscles [120]. However, the sequence of internal laryngeal muscle activation seems invariant relative to the submental complex (mylohyoid, geniohyoid, and anterior digastric muscles) activation, regardless of upright or supine body position [5]. Moreover, submental muscle activity somehow depends on the bolus volume [98]. Extrinsic laryngeal muscles, mainly the supra- and infrahyoid muscle groups, also play a role in phonation. Elevation and depression of the hyoid bone was found to be related to the fundamental frequency of phonation [133]. Changes in the electrical activity of these muscles were observed when singing with different pitches[1]. In a study involving squirrel monkeys, infrahyoid muscle activations were found to be present during various vocalizations [61].  2.3  Swallowing and Aging  Even in healthy elderly people without dysphagia, there is an increased frequency of abnormalities such as sensorimotor incoordination and pharyngeal retention during the swallowing process as compared with younger people [30]. The frequency of laryngeal penetration was observed to be more than double for people over the age of 50 [22]. While elderly people exhibit more frequent swallowing dysfunction, Ekberg and Feinberg [30] suggested that these dysfunctions may not be abnormal in the elderly due to the effect of normal aging. A number of physiological changes in swallowing function occur with age. In general, the duration of various swallowing motions is prolonged, often with a significant delay, and the range of motion is reduced with increasing age. A significantly increased overall duration of oropharyngeal swallowing has been observed in many studies [19, 103, 105, 117, 129, 139]. There is also an increased delay in triggering of 13  2.4. Muscle Aging pharyngeal swallow [70, 73, 129] as well as triggering of UES opening [111]. When looking specifically at hyoid motion, hyolaryngeal excursion is delayed with age [105, 111]. However, the age effects on the range of hyolaryngeal motion have been inconsistent across various studies. Lundy et al. [73], Kern et al. [59], and Yabunaka et al. [139] reported reduced hyolaryngeal excursion in the elderly. In the studies of Logemann et al. [70, 71], maximum vertical and anterior hyoid motion decreased in men but increased in women with age. It was suggested that women maintain better muscular reserve than men. Other than delay, Shaw et al. [111] found age did not affect hyolaryngeal motion during swallowing when examining laryngeal excursion and geniohyoid muscle length, which was used as a measure of hyoid motion. Conversely, Yabunaka et al. [138] showed the extent and duration of geniohyoid muscle motion during swallowing increased gradually with age. Similar results were obtained by Aydoğdu et al. [4], who found that the electrical activity of submental muscles increased in duration and decreased in amplitude. An interesting finding by Sonies et al. [117] was that older subjects exhibited multiple hyoid gestures, sometimes described as tongue pumping, and required more time to accomplish hyoid excursion. However, the relationship between hyoid gestures and range of motion is yet undetermined. UES opening has also been reported to decline significantly in the elderly [59, 111], which may correspond to reduced hyoid excursion. Other changes in swallowing with respect to aging include increased oral transit time [19], longer cricopharyngeal opening [63], and decreased peristaltic velocity [129] to name a few.  2.4  Muscle Aging  Age-related physiological changes in skeletal muscle have been a popular topic of research for several decades. Some of the changes include reduction in muscle strength, loss of muscle mass (atrophy), and prolonged contraction time with advancing age [26, 101]. For the purpose of this thesis, the decline in muscle strength is of most concern. Several studies have attempted to understand and explain this phenomenon by associating atrophy [39, 67], 14  2.4. Muscle Aging loss of motor units [15], and reduction in cross-sectional area [33, 36] with muscle strength reduction. Considering the rate of strength reduction was found to be much more rapid compared with the loss of muscle mass, Goodpaster et al. [39] suggested that there was also a decline in muscle quality, that is, a loss of strength per unit muscle mass. Lynch et al. [74] indeed found age-related muscle quality reduction, but its magnitude depended on the muscle and the type of action (concentric or eccentric) studied. In a 12-year study by Frontera et al. [36], loss in cross-sectional area of 12.5%– 16.1% was found to be significant and could be a major contributor to the reduced muscle strength. Loss of motor units was observed in muscles with advancing age, particularly in people over 60 [15, 26]. However, the remaining motor units often enlarged to partially compensate for the loss. Muscle atrophy and decreases in cross-sectional area could both be caused by a loss of muscle fibres [65, 67]. Older people have 23%–35% smaller limb muscles and significantly more fat and connective tissues than younger people [33]. Muscle strength reduction also seemed to correlate with type II (fast-twitch) fiber area [65]. This could explain why the swallow motions reduce and the durations prolong with advancing age. Grip strength as well as strength of various muscles has been examined to quantify the rate of muscle strength decline. The grip strength of people in their 60s was found to be almost half that of people in their 20s [13]. Kallman et al. [56] found that grip strength declined at an accelerating rate after age 40 and was strongly correlated with muscle mass. Limb muscles were often used in studies on muscle strength. The reduction in muscle strength with age tends to be curvilinear, with relatively stable characteristics in the first few decades and an increasing rate of decline in the later stages. Muscle strength normally peaks in the 20s or 30s. Regardless of magnitude, this trend has been rather consistent across studies [65, 68, 74, 80]. Reduction in muscle strength normally becomes apparent after age 60 [26]. On average, people in their 60s and 70s exhibit a 20%–40% decline in muscle strength [101]. Frontera et al. [36] reported a reduction in muscle strength of 0.75%–2.45% per year, while Goodpaster et al. [39] reported 2.8%–3.6% per year and Skelton et al. [115] reported 1%–2% per year. Lynch et al. 15  2.5. Existing Computer Models [74] generalized the muscle strength decline as 12%–14% per decade after age 50. However, this rate of decline does alter with gender and possibly race. Women tend to have a significantly slower rate of decline compared with men [39]. This supports the argument by Logemann et al. [71] that women, unlike men, maintain better muscular reserve and thus are able to maintain hyoid motion displacement for swallowing in old age. The influence of ethnicity was examined by Goodpaster et al. [39], who found a faster strength decline in African American compared with Caucasian subjects of both genders. However, no significant differences were observed between Caucasian and Hispanic women [106].  2.5  Existing Computer Models  Computer modeling of the anatomical systems largely relies on the available medical imaging techniques. For the hyolaryngeal complex, in particular, which is highly complicated, acquisition of accurate geometry both statically and dynamically is crucial for both construction and validation of the computer model. Traditionally, hyoid motion was studied using VF. Kellen et al. [57] even developed a semi-automatic method for tracking the hyoid bone in VF. It was not until recently that high-quality MRI and computed tomography (CT) began to be used to study swallowing function [10, 37, 46, 52]. The dimensions of the laryngeal framework were reported by [28, 50, 108, 127] based on cadaver or MRI data, which could be used to construct a generic model. However, these studies only include measurements of the cartilages and not of soft structures. The existing computer models of the larynx often focus on the vocal fold. De Vries et al. [24], Gunter [40], and Hunter et al. [51] created deformable models of the vocal fold using FEM to studying its behavior for linguistic research. As for swallowing research, a 3D FEM model of a simplified upper airway was implemented to simulate bolus swallowing [82, 119]. However, there exist very few biomechanical models for simulating hyoid motion. Pelteret and Reddy [97] presented a FEM model that included the tongue, hyoid, and surrounding soft tissues. The suprahyoid muscle 16  2.5. Existing Computer Models group was incorporated, but the study focused on the tongue rather than hyoid excursion. A model for studying the interaction between the intrinsic and extrinsic laryngeal muscles was created by Kob and Butenweg [62], though no results obtained with the model were provided. Hannam et al. [42] specifically studied jaw and hyoid movements during chewing using a dynamic model that included hyoid muscle groups (Figure 2.4(a)). A more sophisticated model that included the face and tongue was also presented by the same group [121] (Figure 2.4(b)). The hyoid muscles were each modeled as a single point-to-point spring, which is a distinct difference from their previous work and the hyolaryngeal model presented in this thesis. Finally, there also exist non-dynamic models of the upper airway created for visualization purposes only [16, 48, 55, 83].  (a)  (b)  Figure 2.4: Examples of existing upper airway models. (a) A jaw-hyoid model by [42] and (b) a face-jaw-tongue model by [121], by permission.  17  Chapter 3  Model of the Hyolaryngeal Complex The goal of this work is to investigative the effects of changes in muscle properties on laryngeal movement. Specifically, the present study explores limitations in the range of hyoid motion as a result of muscle aging. As mentioned in earlier chapters, in vivo studies on swallowing are highly limited by a number of technical difficulties. Therefore, as an alternative to invasive methods for the study of hyoid movements that are specific to a limited set of subjects, here a generic 3D model is described to perform dynamic simulations that can closely mimic the biomechanics of the structures of interest (see Figure 3.1). The resulting model can then be used not only to study the effects of muscle aging on hyoid motion but also to gain a better understanding of the biomechanics related to swallowing and oropharyngeal dysphagia. The hyoid bone is unique in that it is the only bone in the human body that is not articulated directly to any other bones. It is anchored by several pairs of muscles and ligaments that provide close coupling between and to surrounding structures. Some muscles attach to close structures such as the mandible and thyroid cartilage, whereas others extend further and insert onto the sternum and even the scapula. This model incorporates most of the anatomical structures responsible for hyoid maneuvers, such as the supra- and infrahyoid muscle groups as well as tissues that provide structural support to the hyolaryngeal complex, such as the pharyngeal constrictors. Structures such as the mandible and sternum are included to provide attachment sites for the soft tissues.  18  Chapter 3. Model of the Hyolaryngeal Complex  (a) Lateral view  (b) Anterior view  (c) Oblique view  Figure 3.1: The 3D model of the hyolaryngeal complex.  19  3.1. Bones and Cartilages The anatomical complexity of these structures enables the hyolaryngeal complex to perform intricate and highly coordinated movements such as those required for swallowing. In order to closely simulate this behavior, it is necessary for the model to incorporate the different characteristics of each of the structures that can affect the swallowing motion. FEM methods are well established for modeling biological tissues with a wide range of material characteristics. However, FEM simulations of highly discretized models are computationally intensive. The generation of a well-conditioned finite element mesh for geometrically complex structures has also been widely considered a challenging problem [7, 45]. To avoid unnecessary complexity in our model, alternative methods of modeling were utilized where FEM is not crucial. Stiff structures that undergo little deformation, such as bones and some of the cartilages, are modeled as rigid bodies in which no deformation is allowed. Point-to-point springs are used to model muscles and ligaments that do not directly contribute to hyoid movements but provide structural support to the hyolaryngeal complex. This thesis provides details on all the anatomical structures included in this model and how they are represented. Appendix A provides some background knowledge of the anatomy and physiology involved in swallowing. In addition, see Appendix B for the mechanical parameters used in the model.  3.1  Bones and Cartilages  Nine cartilages, three paired and three unpaired, together with the hyoid bone form the laryngeal skeleton (see Section A.1.2). Among these, the thyroid and cricoid cartilages are the largest and make up the bulk of the larynx. Elevation of the larynx is mainly caused by hyoid muscle contractions. The tight coupling between hyoid bone and the two cartilages allows the larynx to move vertically together with the hyoid bone almost as a single body during speech and swallowing. These three components are connected with other structures in the head and neck region with extrinsic muscles of the larynx, which are the focus of this work. Conversely, the other cartilages, with the exception of the epiglottis, are much smaller in size and are 20  3.1. Bones and Cartilages mainly for supporting the internal structure of the larynx, i.e., the vocal folds. Since the purpose of this model is to simulate hyoid movement, the internal biomechanics of the larynx can be neglected; thus, all the intrinsic structures of the larynx are not included as part of our model. Although the epiglottis plays an important role in airway protection and is closely coupled with the hyoid bone, thyroid cartilage, and tongue, the elastic cartilage itself does not have a direct impact on the motion of the hyoid bone, and therefore is not included in the model. The model also includes structures in the head and neck region that are attached by extrinsic muscles of the larynx. Superior structures include the mandible and temporal bone of the skull; each provides the origin for three and two pairs of suprahyoid muscles, respectively. The sternum, clavicles, and scapulas function as origins for the infrahyoid muscles. The upper portion of the vertebrae was included for the sole purpose of providing a visual reference in the 3D simulation. The upper part of the trachea was also included since it connects with the cricoid cartilage and helps stabilize the orientation of the larynx. While human bones do have some degree of elasticity, they only experience relatively minimal deformation under the circumstances of our interest. The same applies to the thyroid and cricoid cartilages. Therefore, all the bones and cartilages in this model were implemented as rigid bodies defined solely by their surface geometries and masses (see Figure 3.2 for the rigid structures in the model). No deformations were allowed and no material properties were associated other than density. Furthermore, except components of the hyolaryngeal complex, whose motions are of interest, external components such as the mandible and sternum were kept stationary. In oropharyngeal swallowing, those structures were commonly observed to be almost motionless and therefore can be modeled as fixed objects during the simulations. Table 3.1 shows a list of bones and cartilages modeled as dynamic and fixed rigid bodies in our model. Some of the non-dynamic structures are denoted as ”partial” because their geometries in the model are incomplete. To minimize unnecessary effort, segmentation was avoided on geometries that were highly complicated and not influential to hyolaryn21  3.2. Muscles geal motion. For example, the base of the cranium is absent in the model. The parameters assigned to each rigid body can be found in Section B.1.  Figure 3.2: Rigid bodies in the model. Bones are shown in gray and cartilages are in yellow.  3.2  Muscles  Since the focus of this work is on the mobility of the hyoid, muscles responsible for moving this bone need to be carefully modeled. The movements of the larynx during speech and swallowing are closely controlled by numerous muscles that provide the degrees of freedom and stability to the structure. The hyoid alone is attached to ten pairs of muscles that originate from multiple directions (see Appendix A). Although each of the muscles  22  3.2. Muscles Name  Type  Pair  Cricoid Hyoid Thyroid Tracheal Rings Clavicle Cranium Humerus Mandible Maxilla Scapula Sternum Vertebrae  Cartilage Bone Cartilage Cartilage Bone Bone Bone Bone Bone Bone Bone Bone  No No No No Yes No Yes No No Yes No No  Modeling Method Rigid body Rigid body Rigid body Rigid body Rigid body Rigid body Rigid body Rigid body Rigid body Rigid body Rigid body Rigid body  Dynamic Yes Yes Yes Yes No No No No No No No No  First 10 rings Partial Partial Partial Partial Partial Partial  Table 3.1: Rigid structures in the model.  plays a certain role in hyoid and laryngeal maneuvers, the hyoglossus was excluded from the model because its main function is to depress and retract the tongue rather than to move the hyoid. More importantly, the hyoglossus muscle is attached to the tongue, which is a muscular hydrostat capable of complex shape changes. The assumption that the hyoglossus muscle has little influence on hyoid movement precludes the need for implementation of the muscular hydrostat, thus simplifying our modeling problem. Table 3.2 shows a list of extrinsic muscles associated with the larynx that is included in the model. In ArtiSynth, a muscle can be implemented either as a finite element mesh or a point-to-point spring (see Figure 3.3 for an example). One of the main advantages of FEM over a simple spring is that it allows us to model muscular tissue as hyperelastic material with nonlinear mechanical behavior, both active and passive. In our model, the active component was modeled by overlaying point-to-point springs on top of the FEM meshes, which will be covered in more detail in Chapter 4. Further, the biomechanics of a geometrically complicated system such as the hyolaryngeal complex is largely defined by its distinctive shape. FEM can compute the deformation of an arbitrary geometry with boundary constraints, which is essential for accurately simulating the interaction between muscles and the surrounding 23  3.2. Muscles  Name Digastric anterior Digastric posterior Geniohyoid Mylohyoid Omohyoid inferior Omohyoid superior Sternohyoid Sternothyroid Thyrohyoid Pharyngeal constrictor, inferior Pharyngeal constrictor, middle Stylohyoid  Pair Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes  Modeling Method FEM FEM FEM FEM FEM FEM FEM FEM FEM Network of passive springs Network of passive springs Active spring  Table 3.2: Muscles in the model.  Figure 3.3: Examples of FEM and spring muscle models. The thyrohyoid muscles were modeled as finite element meshes, whereas the stylohyoid muscles were modeled as springs.  24  3.2. Muscles rigid structures. However, FEM requires a shape to be discretized into a volumetric mesh with a finite set of polyhedral elements; the generation of a volumetric mesh for irregular shapes can be challenging. Thus, in cases where • the muscle’s mass and deformation do not have direct influence on hyoid movement, • the muscle’s deformation can be represented simply as shortening between the origin and insertion point, or • the muscle’s active contraction behavior is not of interest to this project, a point-to-point spring was used in place of a mesh. Table 3.2 indicates which technique was applied to each muscle. Most of the muscles were implemented as FEM meshes since they are directly involved with producing hyoid motion. For the functions of each muscle, see Section A.1.3. One exception is the stylohyoid muscle. Although the contraction of this muscle results in elevation of the hyoid, the shape makes it more appropriate and effective to be modeled as a spring rather than a mesh. All FEM muscles in our model were modeled as Mooney–Rivlin materials with the same parameters as those used by [11, 121] (see Section B.1 for details). Other muscles modeled as point-to-point springs were the pharyngeal constrictors. While not directly involved with the elevation and depression of the larynx, they do provide some support and stability to the larynx. The pharyngeal constructors have three segments and are thin and broad in shape (see Section A.1.3). The middle and inferior segments arise from two sides of the hyoid, thyroid, and cricoid cartilage and insert at the pharyngeal raphe with the opposite side muscle. Due to the unique shapes, utilization of FEM is troublesome. Therefore, a network of springs was used so the larynx can be stabilized under some restrictive force from the posterior direction (see Figure 3.4). Passive springs were used due to the lack of data on the mechanical characteristics of the constrictor muscles. The superior pharyngeal constrictor has no connection with the hyoid mechanically and 25  3.3. Ligaments and membranes  (a) Lateral view  (b) Posterior view  Figure 3.4: Pharyngeal constrictors modeled as spring networks.  thus was excluded from our model. Illustrations of all muscles discussed can be found in Section A.1.3.  3.3  Ligaments and membranes  Other than muscle, soft tissues, including ligaments and membranes, also have significant effects on the dynamics of the hyolaryngeal complex. The laryngeal skeleton is held together by a number of connective tissues. Tissues internal to the larynx such as the vocal ligaments were all excluded from the model for they do not have an effect on the overall laryngeal movement. There are in total four ligaments/membranes that are important for our purpose. The thyrohyoid membrane connects the hyoid bone and thyroid cartilage, and therefore, gives the larynx its tight coupling to the hyoid. The cricothyroid ligament connects the thyroid cartilage to the cricoid cartilage, and the cricotracheal ligament connects the cricoid cartilage to the first ring of the tracheal. The annular ligaments of the trachea are a group of circular ligaments that connect the tracheal rings with each other. See Section A.1.2 for details on these ligaments. Another type of connective tissue in our  26  3.3. Ligaments and membranes  (a) Anterior view  (b) Lateral view  (c) Posterior view  Figure 3.5: Connective tissues in the model.  (a)  (b) Zoomed in view  Figure 3.6: Omohyoid intermediate tendon sheath modeled with a FEM mesh and a network of springs.  27  3.3. Ligaments and membranes model includes the tendons that connect two muscle bellies. The omohyoid and digastric muscles both have such intermediate tendons. Similar to muscles, connective tissues in our model were implemented in one of the following two ways: FEM or point-to-point spring. The difference between muscles and connective tissue models is that the former include an active component for generating contractile force. Mechanical properties that reflect the behavior of ligaments and tendons were also applied. Connective tissues were modeled as linear materials in FEM, where suitable Young’s modulus and Poisson’s ratio values were assigned (see Section B.2). Table 3.3 shows a list of connective tissues and their implementation in our model. Name Annular ligaments of trachea Cricothyroid Cricotracheal Thyrohyoid Digastric Omohyoid  Type Ligament Ligament Ligament Membrane Intermediate tendon Intermediate tendon  Pair No No No No Yes Yes  Modeling Method Network of springs FEM Network of springs FEM FEM connected to a network of springs FEM connected to a network of springs  Table 3.3: Connective tissues in the model. Note that all springs listed here are passive.  The cricotracheal ligament and annular ligaments of the trachea were modeled as a spring network that runs between the attachment sites of these ligaments, as shown in Figure 3.5. The purpose of this implementation was to use the network of springs as a simplified representation of the flexible, yet retrained, mechanical coupling that the trachea has on the hyolaryngeal complex. The two intermediate tendons were another special case; they are both sheathed by fibrous tissues that hold them in position. This connection is simplified by the use of a FEM mesh to represent the tendon body and a ring-shaped spring network to represent the constraints that result from the fibrous sheaths (see Figure 3.6 for an example). The parameters associated are provided in Section B.4.  28  3.4. Joint  3.4  Joint  The larynx is considered to have two joints: the cricoarytenoid and cricothyroid articulation. The former connects the cricoid cartilage with the arytenoid cartilage and does not play a role in laryngeal elevation. The later is a joint that connects the cricoid and thyroid cartilage. This joint constrains the inferior horn of thyroid cartilage to the sides of cricoid cartilage and allows rotation of the thyroid cartilage with respect to the cricoid cartilage. In the model, minor translational movement, which is allowed in real human anatomy, was disregarded and the cricothyroid joint was represented as a revolute joint with a transverse axis that runs between the inferior horns of the thyroid cartilage (Figure 3.7).  (a) Posterior view  (b) Lateral view  Figure 3.7: Cricothyroid articulation modeled as a revolute joint.  29  Chapter 4  Implementation of the Model This chapter presents the process flow taken for creating the biomechanical model of the hyolaryngeal complex described in Chapter 3. Note that the process is specific to dynamic modeling in ArtiSynth, an open-source 3D dynamic simulation platform [34]. Other software like Amira and UBCReg [12] are used for processing the geometric data. Through various steps, this process flow takes a set of medical images and transforms them into a 3D biomechanical model. Figure 4.1 shows a block diagram of this flow. While the process flow was developed for the purpose of hyolaryngeal modeling, it can also be applied for creating general-purpose biomechanical models of other musculoskeletal regions.  Figure 4.1: Process flow of biomechanical model generation.  In our method, the generation of a 3D biomechanical model involves several steps: segmentation, symmetry making, volumetric mesh fitting, muscle attachment, muscle fiber specification, coupling, and finally mechanical properties specification. Each step is explained in detail in the following sections. 30  4.1. Data Source and Segmentation  4.1  Data Source and Segmentation  The selection of image data is important, especially for modeling a highly geometrically complex system such as the upper airway. Soft tissues in the hyolaryngeal complex are compact and small in size, which makes details of these tissues very difficult to capture using existing imaging techniques. Even with state-of-the-art MRI techniques, clear visible borders between muscles that lie adjacent to each other, such as the infrahyoid muscle group, cannot be easily found. Although MRI does provide outstanding contrast across various materials as compared to CT, imaging of the thin line of separation between the same types of tissue is still challenging. The Visible Human Project c (VHP) acquired data from representative male and female cadavers in the form of CT, MRI, and cryosection images. Among these data sets, the CT and cryosection images of the male cadaver were selected for our model. The main reason for using cryosection images that they offer much higher visibility of the boundaries between soft tissues, specifically muscles, compared with the other imaging modality. This is necessary for obtaining precise geometries of the hyoid muscles. CT data is also used alongside the cryosection images for their high spatial resolution and bone-to-soft-tissue contrast. Another advantage of the VHP data is that the images were taken from a single subject at the exact same position, which makes image registration between data sets much easier. Manual segmentation is performed on the cryosection data to acquire geometrical information of the soft tissues (Figure 4.2(a)). The cross-sectional area of each muscle is manually labeled throughout the axial image slices (see Figure 4.3 for an example). Since it is difficult to distinguish ligaments and membranes from the surrounding tissues such as fat or other connective tissues, their geometries are not obtained via segmentation. Tendons that run between a muscle and their attachment sites are also omitted in the process of segmentation due to their low visibility in the cryosection images. Details on how the connective tissues are incorporated in the model can be found in Section 4.6. Geometries of the bones and cartilages are obtained from the CT data 31  4.1. Data Source and Segmentation  (a)  (b)  Figure 4.2: Surface geometries segmented from VHP datasets. (a) Soft tissue segmented from cryosection, (b) rigid structure segmented from CT.  Figure 4.3: Labeling of some VHP cryosection images.  32  4.2. Making Symmetry first using threshold segmentation (Figure 4.2(b)). Due to the high contrast difference between the rigid structures and soft tissues, the bones and cartilages can easily be extracted from the CT images. Once they are separated from the rest of the materials, some manual intervention is required. Boundaries between individual components need to be specified because there are surfaces in contact between structures like the mandible and maxilla, which cannot be distinguished in an automated fashion. Furthermore, metallic implants in the body such as dental fillings can distort CT images. In the case of the VHP male subject, regions of the teeth and mandible need to be manually cleaned. Note that the tracheal rings were the only rigid structures whose geometries were not obtained from the image dataset. The shape of a tracheal ring was artistically constructed from dimensions loosely based on [28]. Procedures, including hole filling and smoothing, were applied to both labeled data. Surface meshes of the 3D geometries were then generated from the labeled fields, as shown in Figure 4.3. Note that a raw surface mesh is composed of a large set of triangles, which is computationally intensive to process and not essential for our purpose. Hence, simplification was also carried out to reduce the number of triangles for each mesh. All procedures described in this section were performed with the commercial 3D visualization software package Amira c .  4.2  Making Symmetry  Although the human body lacks complete bilateral symmetry, the anatomical structures of the head and neck region do show a level of symmetry across the mid-sagittal plane when looking at individual/paired structures. This symmetry is never exact and can be caused by genetic or environmental factors such as mechanical loading [3, 87]. For example, studies have shown that the asymmetry observed in human humerus reflects the hand preference of the individual [21, 110]. Asymmetry of the segmented surface meshes is also dependent on the head and body positioning of the imaged subject. Slight misalignment of the head and body could affect the plane of symmetry for a number of structures differently. 33  4.2. Making Symmetry For the purpose of creating a generic hyolaryngeal model and to avoid complication during the simulation, the surface meshes must undergo a process for ensuring bilateral symmetry. Since all muscles of interest in our model are paired, symmetry can be easily achieved by mirroring. However, most human structures do not exhibit exact symmetry and the line of reflection can be inconsistent between regions. Therefore, a simple technique is described for accomplishing bilateral symmetry though an iterative process of morphing and averaging. UBCReg, a software program provided by the TIMC-IMAG Laboratory based on the work by Bucki et al. [12], was used to morph a source surface mesh into the target surface mesh. Each step of this method is described here along with graphical illustrations using the 2D example in Figure 4.4: 1. Given an asymmetric surface mesh Si (Figure 4.4(a)), reflect Si with respect to the mid-sagittal plane to obtain SiR (Figure 4.4(b)). 2. Morph Si to SiR using UBCReg to obtain SiM (Figure 4.4(c)). 3. For each respective node nj in Si and SiM (Figure 4.4(d)), compute the midpoint (Figure 4.4(e)), S( i + 1)[nj ] =  Si [nj ] + SiM [nj ] 2  (4.1)  4. Let i = i + 1 and repeat from step 1; see Figure 4.4 (f)-(h). To achieve an acceptable level of symmetry, the above steps can be repeated several times. Note that this method can be applied under the assumption that there are only minor discrepancies between the original and reflection. This is a reasonable assumption for human anatomical structures when the plane of reflection is the mid-sagittal plane. As one would expect, the number of nodes in the input surface mesh significantly affects the quality of the resulting output mesh produced by this method. This is partly because the performance of the morphing algorithm relies on the level of detail the two meshes provide. Node distribution is 34  4.2. Making Symmetry  (a)  (b)  (c)  (d)  (e)  (f)  (g)  (h)  Figure 4.4: Illustration of the symmetry making process of a shape (a) and its reflection(b). (c) The original shape is morphed into its reflection. (d) The reflection and the morphed shape. (e) Compute mid-points. (f) Result of first iteration. (g) Reflection of (f). (h) Result of the second iteration. 35  4.2. Making Symmetry another factor that can affect the output quality. If the nodes are unevenly distributed across the surface, it is likely that some regions cannot achieve true symmetry even after many iterations. Generally speaking, better result can be expected from this method when using a shape defined by a larger set of triangles compared with the same shape defined by a smaller set of triangles. While this method does provide geometrical symmetry to a surface mesh on a higher level, exact symmetry cannot be achieved (see Figure 4.5 for examples). However, once the shapes are consistent and the planes of reflection for all meshes are aligned, meshes can then be mirrored to obtain a nodeto-node level of symmetry. For paired structures such as muscles and some of the bones, mirroring is easily achieved by reflecting structures from a selected side. However, for structures centered at the mid-sagittal plane, such as the mandible, mirroring requires the surface mesh to be sliced in half, reflected, and then stitched together. For our study, such precise symmetry is not essential, and therefore, the mirroring step was omitted for unpaired structures. To provide an overview of the entire symmetry making process, Figure 4.6 shows the muscles and cartilages of our model at each stage of the procedure. The first image shows the original surface meshes obtained from segmentation in grey and their reflections in purple. As mentioned earlier, there are inconsistencies and misalignment across regions. There exist differences in not only angle but also direction of rotation between a structure and its reflection. For example, there is a rotation about the longitudinal axis for the thyroid cartilage, whereas a rotation about the coronal axis can be observed for the digastric posterior muscle. The second image shows the result obtained from the iterative morphing method. Here, minor discrepancies can be found between nodes but the general shapes retain a level symmetry with respect to the mid-sagittal plane. After mirroring, the last image shows the final result with exact node-level symmetry.  36  4.2. Making Symmetry  (a) Sternohyoid muscle  (b) Thyroid cartilage  Figure 4.5: Example of making geometry symmetric. First image in (a) shows the left sternohyoid in green and the reflected right sternohyoid in red, whereas the second image shows the same shapes made symmetrical using our procedure. (b) shows the thyroid cartilage in green together with its own reflection in red. Note that the symmetries are not at face-level.  37  4.2. Making Symmetry  (a) Original  (b) Morphed and averaged  (c) Mirrored final  Figure 4.6: The process of making the model symmetric.  38  4.3. Volumetric Mesh Fitting  4.3  Volumetric Mesh Fitting  Once the desired surface meshes are obtained, it is necessary to generate volumetric meshes for modeling the structures as finite elements. A volumetric mesh is essentially a volume discretized by a set of polyhedrons. The types of polyhedrons commonly used in FEM analysis are tetrahedral and hexahedral. Figure 4.7 shows the thyrohyoid and mylohyoid muscles of both types. There are advantages and disadvantages to both types of meshes. One main reason people use a tetrahedral mesh is because a tetrahedral mesh can be automatically produced using meshing software such as TetGen, which was developed by Si [113]. Amira c also has a built-in feature for generating a tetrahedral mesh.  (a) Tetrahedral mesh  (b) Hexahedral mesh  Figure 4.7: Examples of volumetric mesh.  However, the generation of a hexahedral mesh is algorithmically more challenging. Su et al. [124] categorized existing approaches for hexahedral meshing into three genres: block decomposition, advancing front, and su39  4.3. Volumetric Mesh Fitting perposition. In the block decomposition approach, the input geometry is divided into smaller sections in which meshes can be generated using various discretization methods. The work of Sheffer et al. [112] is an example of this approach, where the embedded Voronoi graph is first used to divide the object, then each section is individually meshed and finally merged and smoothed together. The advancing front approach progressively constructs hexahedral elements from the surface toward the center of the object. The Whisker Weaving algorithm [126] is a classic example of this approach for generating an unstructured all-hexahedral mesh. Owen and Saigal [89] also use an advancing front technique, but instead the hexahedral elements are derived from an initial tetrahedral mesh. The last one is the superposition approach, also known as grid-based mesh generation. This approach involves the use of a predefined grid that can either be transformed or sculpted to produce an initial mesh. Various techniques can then be applied to fit it to the boundaries of the object. In the algorithm proposed by Schneiders [107], a regular grid is first used to fill the interior of the object. Subspaces between the grid and the boundary are then meshed simply by filling them with quadrilaterals in each 2D cross-section of the grid interval. While there exist several hexahedral meshing techniques that claim to be automatic, current commercially available meshing tools such as the meshing feature in ANSYS R generally require a certain level of manual effort. This might due to two reasons. First, automatic methods may not be able to produce well-condition meshes for all arbitrary shapes. Special cases such as holes could also be problematic. Secondly, it is difficult to generate a structured mesh automatically without any additional information other than the shape, especially for complex geometries. While the meshing process is much more complicated, a hexahedral mesh does provide several benefits over a tetrahedral mesh. It has been reported that tetrahedral elements experience volumetric locking when simulating incompressible material [49]. Biological tissues, including muscle, are nearly incompressible [76, 88, 100], which makes a hexahedral mesh more preferable. Moreover, a smaller number of elements are generally required when using hexahedrons to represent a shape. Further, unlike tetrahedral ele40  4.3. Volumetric Mesh Fitting ments, a hexahedral mesh can be structured, which is more illustrative and computationally efficient due to the simple connectivity of the elements. In terms of accuracy, conflicting results have been reported by researchers when comparing these two types of meshes. Benzley et al. [6] found that a hexahedral mesh produces a more accurate solution compared with a tetrahedral mesh, whereas Nesme et al. [86] found little difference between the two. Ramos and Simoes [104] even reported that the results produced by a tetrahedral mesh are closer to the theoretical results as compared with the results produced by a hexahedral mesh when modeling the proximal femur. In our model, hexahedral elements are utilized for the many advantages listed above. While mesh generation for arbitrary shapes is complicated, the objects in our model that require volumetric meshing are very geometrically specific. By taking into account the simplicity of individual muscle shapes, this study implemented a template-fitting technique that has a limited application to generating a hexahedral mesh for beam- and cylindrical-like shapes. This is similar to the meshing technique utilized by Blemker and Delp [8]. This method involves the use a template mesh, either a beam or a cylinder. The template is mapped to fit the shape of the input surface mesh though a series of projections.  4.3.1  Templates  To accommodate muscles with circular and rectangular cross sections, two parametric templates are generated: beam and cylinder. The beam template is simply made of a regular grid of hexahedrons (see Figure 4.8(a)). It is parametrically defined by the number of elements on each of the three axes. The cylinder template is constructed from a set of beam templates (see Figure 4.8). Five beam meshes are first used to form a cross shape. The side nodes are then merged and every node is readjusted so they are distributed radially from the center. Three parameters are also used to specify the number of elements in the cylinder template. Intuitively, the quality of result mesh greatly depends on the geometric similarity between the template and the input surface mesh as well as the resolution of the  41  4.3. Volumetric Mesh Fitting template.  (a) Beam template  (b)  (c)  (d) Cylinder template  Figure 4.8: Two types of mesh templates. Cylinder template (d) is constructed from (a) though the process of (b) and (c).  4.3.2  Fitting  Once a template is selected, the elements need to be fitted into the object volume. This procedure can be decomposed into a series of 2D operations. The template is processed interval by interval. The set of nodes at each cross-sectional interval along the principal axis are fitted to surface objects independently. The procedures of this fitting technique are detailed below, along with illustrative drawings in Figure 4.9: 1. Given a surface object, compute its principal axis. 42  4.3. Volumetric Mesh Fitting 2. Denote the principal axis as the z-axis and compute the bounding box. 3. Given a user defined margin zmargin , discretize the bounding box along the z-axis into nz number of intervals, where nz defines the number of elements along the principal axis. 4. At every interval, fit the corresponding nodes to the 2D cross section of the surface object. The method of fitting depends on the type of template selected. 5. Finally, project the nodes on the first and last interval onto the boundary of the surface object along the positive and negative z-axis directions. Based on the template selected, different 2D fitting strategies are applied. Both strategies use ray casting for placing nodes on the surface of the object. The main difference is that ray casting is done orthogonally with respect to the axes in the case of a beam template and radially with respect to the center of the surface object cross section in the case of a cylinder template. In the following section, the details of both fitting strategies are described in detail. Note that ray casting operation is denoted as point, direction where it represents the origin point and the direction of the ray to be casted. ~x and ~y are used to denote the unit vectors in the x and y directions, respectively. The procedure for fitting a rectangular grid to a surface cross section is described here with illustrative drawings shown in Figure 4.10: 1. Given a cross section and user-defined margins, xmargin and ymargin , the bounding box is defined by the four corner points e0−3 as shown in Figure 4.10(a). 2. To find the upper-right corner node, perform ray casting on {e0 − ymargin × ~y , −~x} 3. If the intersection does not fall in the effective area, as shown in green in Figure 4.10(b), repeat the ray casting process on {e0 − (ymargin + i × step) × ~y , −~x} with an incrementing i. See Figure 4.10(c). 43  4.3. Volumetric Mesh Fitting  (a)  (b)  (c)  (d)  Figure 4.9: Graphical illustration of the template fitting technique: first four steps.  44  4.3. Volumetric Mesh Fitting  (a)  (b)  (c)  (f)  (d)  (g)  (i)  (j)  (e)  (h)  (k)  (l)  Figure 4.10: Illustration of rectangular grid fitting.  45  4.3. Volumetric Mesh Fitting  (a)  (b)  (c)  (d)  (e)  (f)  (g)  (h)  (i)  (j)  Figure 4.11: Illustration of finding the centroid of a 2D shape.  46  4.3. Volumetric Mesh Fitting 4. Perform a similar ray casting operation for the three other corner nodes: {e1 + (ymargin + i × step)~y , −~x} for upper-left corner, {e2 − (ymargin + i × step)~y , −~x} for bottom-right corner, {e3 + (ymargin + i × step)~y , −~x} for bottom-left corner. 5. Once all four corner nodes are placed (Figure 4.10(d)), perform a series of ray casting steps between the corner nodes to place all the boundary nodes as shown in Figure 4.10(e)-(h). The widths of the x and y intervals depend on the user-assigned nx and ny values, which represent the number of elements along the x- and y-axes, respectively. 6. Interpolate the internal nodes based on the boundary nodes of the same column; see Figure 4.10(i)–(j). 7. Finally, all nodes are placed and the results are shown in Figure 4.10(k)– (l). Though the above strategy, a rectangular quadrilateral grid can be fitted into a non-regular 2D shape. By repeating this procedure for every crosssection interval of the 3D surface object, the beam template is fitted to the volume, thus producing an all-hexahedral mesh of the object shape. The procedure for fitting a circular grid to a surface cross section is described here with illustrative drawings shown in Figure 4.11 and 4.12: 1. Given a cross section and an arbitrary point on the plane (Figure 4.11(a)), the centroid of the area first needs to be found. 2. Perform ray casting radially away from the point (Figure 4.11(b)). 3. If the number of intersections detected is less than 1, double the angular resolution of the ray casting operations; see Figure 4.11(c). 4. Compute the centroid of the intersections detected (Figure 4.11(d)). 5. If the number of intersections equals the number of ray castings, double the angular resolution as shown in Figure 4.11(g)–(i). 47  4.3. Volumetric Mesh Fitting 6. Repeat step 2–5 until the Euclidean distance between the previous and current centroid converges (Figure 4.11(j)). 7. Once the centroid is found, perform ray casting radially from the centroid to place the boundary nodes (Figure 4.12(a)). The angular resolution is determined by a user-assigned value nr that represents the number of elements along the circumference of the cylinder template. 8. Interpolate the internal nodes radially based on the boundary nodes; see Figure 4.12(b)–(c). 9. Finally, all nodes are placed and the results are shown in Figure 4.12(d)– (e). Through the above strategy, a circular quadrilateral grid can be fitted into a non-regular 2D shape. By repeating this procedure for every crosssection interval of the 3D surface object, the cylinder template is fitted to the volume, thus producing an all-hexahedral mesh of the object shape.  (a)  (c)  (b)  (d)  (e)  Figure 4.12: Illustration of circular grid fitting.  48  4.3. Volumetric Mesh Fitting  4.3.3  Comparison Between the Two Templates  As mentioned, the quality of the resulting volumetric mesh greatly depends on the geometric similarities between the input shape and the selected template. Figure 4.13 shows an example of the same shape fitted using the two types of grids. The rectangular grid consists of 15 elements, whereas the circular grad consists of 20. In this example, the rectangular grid seems to produce better-conditioned elements without the need to increase element count. Figure 4.14 shows mesh fitting of the two templates on an idealized muscle shape, where the principal axis is shown in red. Due to the circular cross section, the cylinder template produced a hexahedral mesh where the boundary elements are better conditioned. Moreover, the nodes are more uniformly distributed across the volume with rotational symmetry of the muscle preserved, unlike in the beam template result.  (a)  (c)  (b)  (d)  Figure 4.13: Rectangular and circular grid fitting example.  In general, the cylinder template produces meshes in which the quality of elements is consistent throughout the volume. The beam template tends to produce a mesh with very well conditioned internal elements and poorquality corner elements. On the other hand, elements are denser toward the center in the cylinder template. This is undesirable since the surface of an  49  4.3. Volumetric Mesh Fitting  (a)  (b)  (c)  (d)  (e)  (f)  (g)  Figure 4.14: Template fitting of an idealized muscle.  50  4.3. Volumetric Mesh Fitting  (a)  (b)  (c)  (d)  (e)  (f)  (g)  (h)  Figure 4.15: Hexahedral mesh fitting on the digastric anterior muscle using the beam template. 51  4.4. Muscle Attachment  (a)  (b)  Figure 4.16: Example of inverted elements. Digastric anterior and posterior muscle meshes with the inverted element highlighted in red.  object is usually where more detail is required in FEM analysis. In other words, the beam template will likely require fewer elements to represent the same object since it has a higher element resolution at the boundary. Thus, the beam template is used for most of the muscles in the hyolaryngeal model. This is because many of the muscles are flat and have a cross section that is closer to a rectangle than a circle. The cylinder template yields very poor results especially in cases such as the sternohyoid and sternothyroid muscles, where the muscles are thin and wide. Figure 4.15 shows the beam template fitting process for the digastric anterior muscle. While this template fitting technique provides an automatic way of generating hexahedral meshes, it is limited to particular shapes. Moreover, it often provides poor quality and sometimes inverted elements at the two ends of the muscle. An example is shown in Figure 4.16, where the inverted element at the end of the digastric posterior muscle is highlighted in red. These inverted elements need to be fixed by manually readjusting the nodes.  4.4  Muscle Attachment  After a volumetric mesh has been generated for every muscle, the ends of the muscle need to be attached to appropriate origin and insertion sites. 52  4.5. Muscle Fiber Specification As mentioned, tendons that run between a muscle and a rigid structure are not segmented out as part of the muscle volume. Therefore, there is a discontinuity that has to be repaired. This is done by manually relocating the end nodes. Figure 4.17 shows how the end nodes of the sternohyoid are moved to connect them to the hyoid bone. The appropriate attachment sites are approximated by referring to cryosection images along with anatomical drawings from the literature (see Section A.1.3). Once the end nodes are placed in the correct location, they are fixed with the corresponding rigid body in ArtiSynth. Then, the muscles are coupled with the bones and cartilages in the model. Figure 4.18 shows the muscle attachments in our model, where the attached nodes are drawn in blue.  Figure 4.17: Attaching the sternohyoid muscle.  4.5  Muscle Fiber Specification  To model muscles as force actuators, the direction of muscle contraction needs to be specified for each muscle. ArtiSynth supports two types of muscle models. The first type consists of a set of springs superimposed on the FEM mesh, as shown in Figure 4.19(b). These springs are specified by the user by providing the software with a set of nodes or markers that indicate how the springs are connected. Since hexahedral meshes are generated using the template fitting technique along the principal axis, elements follow the muscle direction quite closely, which makes specification of the springs 53  4.5. Muscle Fiber Specification  (a)  (b)  (c)  (d)  (e)  (f)  Figure 4.18: Muscle attachments are shown in blue. (a) Full frontal view. (b) Anterior, (c) oblique and (d) lateral views of the surpahyoid muscle attachments. (e) Anterior view of attachments to the sternum and (f) posterior view of attachments to the hyoid.  54  4.6. Coupling almost effortless. This was the muscle type utilized by our model, and all simulations were run using this model. Details on how the muscle springs were characterized mathematically can be found in Section B.3. Figure 4.20 (a), (c), and (e) show muscles in our model with superimposed spring fibers. The second type of muscle model requires a direction to be associated with each finite element (see Figure 4.19(c)). Artisynth supports transverse anisotropic elements, including constitutive models by Weiss et al. [135] and Blemker et al. [9], allowing muscle force directions to be specified independent of the element node geometry. While assigning a contraction direction to every element is tedious, ArtiSynth comes with a built-in capability that can generate the second type of muscles from the first type. This is done by interpolating element contraction directions from surrounding springs. The element directions in Figure 4.19(c) are generated from the springs in Figure 4.19(b). Figure 4.20 (b), (d), and (f) show muscles with element directions assigned.  (a)  (b) Type 1  (c) Type 2  Figure 4.19: Fiber specification of the thyrohyoid muscle.  4.6  Coupling  This section describes some miscellaneous steps taken specifically for the purpose of modeling the hyolaryngeal complex. They mostly involve simplification of anatomical structures to a more geometrically or mathematically straightforward representation and were accomplished manually. 55  4.6. Coupling  (a) Type 1, anterior view  (b) Type 2, anterior view  (c) Type 1, posterior view  (d) Type 2, posterior view  (e) Type 1, lateral view  (f) Type 2, lateral view  Figure 4.20: Two types of muscle models. Type 1: Muscles with springs superimposed, (a), (c), and (e). Type 2: Muscles with element force directions, (b), (d), and (f).  56  4.6. Coupling  4.6.1  Connective Tissues as FEM Meshes  Structures that were difficult to segment from image data were incorporated in the model by manually creating the volumetric meshes. The geometries and attachment sites were determined from anatomical drawings in the literature. Connective tissues, including ligaments and tendons, were mostly constructed in this way for the present model. Figure 4.21 shows handcreated FEM meshes for the laryngeal ligaments.  (a) Anterior view  (b) Lateral view  Figure 4.21: Ligaments of the larynx.  4.6.2  Active and Passive Springs and Spring Networks  As explained in Chapter 3, there are structures that were better modeled as springs instead of FEM meshes in our model. There are two types of springs in ArtiSynth: passive and active. There are no masses associated with either types. A passive spring is simply defined by its stiffness and damping coefficient, whereas an active spring is characterized by a more elaborate Hill-type formulation that takes muscle contraction force into account [44, 141] (for more details, see Appendix B). Muscles like the stylohyoid muscle were implemented as active springs, whereas the coupling between the larynx and trachea was implemented as 57  4.7. Mechanical Properties Specification a network of passive springs. All the springs were defined by their points of attachment, which were visually estimated. Examples of the spring network implementation can be found in Section 3.3.  4.6.3  Constraints  Finally, joints were implemented as constraints between two rigid bodies. Although ArtiSynth does provide rigid body constraints of various types, a revolute joint was used for the cricothyroid joint, which was the only joint in our model (see Figure 3.7).  4.7  Mechanical Properties Specification  At this point, the model was geometrically complete. This last step involved specification of constitutive models and mechanical properties for each component. Different modeling techniques require different parameters to be addressed. Density or mass is the only property required for rigid bodies. In contrast, FEM and spring components are defined by various mechanical parameters depending on the constitutive model used. See Appendix B for all the mechanical parameters used in the model. Once all parameters were assigned appropriate values, the model was complete and ready for simulation. Figures of the final model can be found in Figure 3.1 in Chapter 3.  4.8  Summary  In this chapter, a process flow for generating a dynamic model of the human hyolaryngeal complex from a set of medical images was presented. The steps include segmentation, symmetry making, volumetric mesh fitting, muscle attachment, muscle fiber specification, coupling, and finally, mechanical properties specification. There are several limitations associated with this model. For one, the human head and neck anatomy exhibits high variability across individuals. Variations in each extrinsic laryngeal muscle were described by [136]. How-  58  4.8. Summary ever, the present model was implemented based on geometrical data from a single subject, which means it might only be able to simulate behaviors specific to subjects with similar morphological characteristics. Moreover, the mechanical properties assigned to the model were mostly approximations based on past studies. Many of the parameters were estimated from generic tissue properties, which are not specific to the structures modeled here. Other parameters were based on educated guesses, which likely have a negative impact on the precision of simulation. Another limitation is that the effects of surrounding structures such as the tongue and deformable pharyngeal wall were neglected in this model, although this does not match reality. Regardless of the above and except for tissues that were simplified to springs (for example, pharyngeal constrictors), the geometries of the model were mostly obtained from segmentation of medical images of a real human subject. By incorporating information from literature sources such as Gray’s Anatomy [136], the model yields a high level of anatomical accuracy. By combining various modeling techniques, i.e., FEM and active and passive spring models, dynamic simulation of this highly complex biomechanical system can finally be realized and functional studies can be performed. Chapter 5 and 6 present studies of hyoid motion utilizing this model.  59  Chapter 5  Evaluation I: Muscle Functions and Swallowing Simulation Once a model has been developed, it is necessary to perform validation against clinically measured data. However, due to intersubject variability and difficulties in measuring the dynamic motion of soft tissues, clinical validation of a hyolaryngeal model could be extremely challenging. The first part of this chapter instead describes the action of each extrinsic laryngeal muscle and compares it to what is known in the current literature. Thus, it provides a functional validation of the model. In the second part of this chapter, the ability of the model to reproduce realistic hyoid motion during swallowing is discussed. A series of inverse simulations were performed using motion data obtained from VFs of swallowing trials, and the corresponding potential muscle excitations were computed. Selective usage of extrinsic laryngeal muscles was also examined.  5.1  Individual Muscle Function Simulation  This section examines how individual muscles contributes to the overall hyoid movement in the developed model. Each muscle pair was individually activated and the resulting hyoid motion was observed. A simple activation waveform, as shown in Figure 5.1, was utilized. All simulations discussed in this chapter were performed with an acceleration due to gravity of 9.8 m/s2 in the z-direction and a maximum time step of 1.0 ms.  60  5.1. Individual Muscle Function Simulation  Figure 5.1: Muscle activation waveform.  On the mid-sagittal plane, a marker was placed at the anterior-inferior corner of the hyoid, which is indicated by the green sphere in Figure 5.2. The marker was used to trace the position of the hyoid. To observe the motion of the hyoid as a result of muscle activations, images of the model in midsagittal view were captured once the activation waveform reached a plateau (at 0.5 s). Figures 5.3–5.5 show the hyoid motion resulting from contractions of each extrinsic laryngeal muscle pair. The green sphere represents the current hyoid position (at 0.5 s), while the blue sphere represents the initial hyoid position (at 0 s). The green line that connects the blue sphere with the green sphere is the trajectory by the hyoid from time 0 to 0.5 s. Note that although only the activated muscle is displayed in the images, the rest of the muscles were not excluded during the simulations. The results obtained from this model are in good agreement with the muscle functions described in the literature. According to anatomy literature [128, 136], suprahyoid muscles are responsible for elevating the hyoid as well as moving it anteriorly toward the mandible. The action of each supra- and infrahyoid muscle is listed in Table A.1. In our simulations, hyoid elevation could be achieved either by contracting the digastric posterior, mylohyoid, or stylohyoid muscles; whereas, anterior movement could be achieved by contracting the digastric anterior or geniohyoid muscles, as shown in Figure 5.3. The main function of the infrahyoid muscles is to depress the hyoid along with the larynx. The sternohyoid, sternothyroid, and 61  5.1. Individual Muscle Function Simulation  Figure 5.2: Mid-sagittal view of the hyoid and larynx at rest position.  omohyoid superior muscles all have the ability to lower the hyoid 5.4. As for the thyrohyoid muscle, it reduces the distance between the hyoid bone and thyroid cartilage (Figure 5.5). Note that in our model, some muscles such as the digastric posterior and sternohyoid muscles can pull the hyoid toward the vertebrae when activated. This effect can be neglected since in real human anatomy, surrounding structures (which are not included in our model) such as the tongue and the esophagus would likely restrain the hyoid from moving too far backward from its resting position. Based on the simulation results of individual muscle contractions, these muscles can be organized into three groups based on their ability to maneuver the hyoid bone: elevation, depression, and anterior motion. The hyoid motion created by each of the three muscle groups are shown in Figure 5.6. In this part of the simulations, all muscles within the group were activated together using the same waveform. The thyrohyoid muscle was excluded in this study since the relative motion between the hyoid bone and thyroid cartilage is not of interest. 62  5.1. Individual Muscle Function Simulation  (a) Digastric anterior  (b) Digastric posterior  (c) Geniohyoid  (d) Mylohyoid  (e) Stylohyoid  Figure 5.3: Actions of the suprahyoid muscles.  63  5.1. Individual Muscle Function Simulation  (a) Omohyoid superior  (b) Sternohyoid  (c) Sternothyroid  Figure 5.4: Actions of the infrahyoid muscles.  (a) At rest  (b) Activated  Figure 5.5: Actions of the thyrohyoid muscles.  64  5.2. Inverse Simulation of Swallowing Motion  (a) Anterior motion  (b) Elevation  (c) Depression  Figure 5.6: Group actions of the extrinsic laryngeal muscles.  5.2  Inverse Simulation of Swallowing Motion  The functions of individual muscle pairs in our model were shown to be in good agreement with the literature in the previous section. To further validate the model, the next step was to demonstrate that the model was capable of mimicking realistic swallowing behavior. In this section, the ability of the model to reproduce hyoid movement using clinically measured data for normal swallows is described and the possible muscle activations are examined. The experiment can be divided into two parts. In the first part, inverse simulations were performed on swallowing data obtained from  65  5.2. Inverse Simulation of Swallowing Motion individual subjects and the results were compared. In the second part, the selective usage of muscles for hyoid excursion was examined by performing inverse simulation using subsets of muscles.  5.2.1  Data Source  VF data of normal swallowing was supplied by Christy Ludlow of the National Institute of Neurological Disorders and Stroke and published in [122]. Three healthy subjects each performed a swallowing trial with 5 ml of water while lateral VF was recorded. Landmark positions were manually selected on each frame of video. Figure 5.7 shows the location of landmarks, and Table 5.1 provides brief descriptions on how they were defined. The coordinate frame was defined by the lower anterior corner of the second (C2) and fourth (C4) vertebrae, where the y-axis was defined as the line between C2 and C4. The landmark positions were measured with respect to the position of C4 on each frame. For the purpose of this evaluation, only the motion of the hyoid was studied. The trajectory of the hyoid during a normal swallowing cycle was obtained from the position of the lower anterior point on the hyoid body (HA) obtained from the VF. Since the trajectories were to be used by the model for simulation, the data needed to be transformed into the coordinate frame of the model. The positions originally measured with respect to C4 were converted into relative displacements with respect to the initial HA position on the first frame of each video. While the coordinate frame of the VF measurements was defined by C2 and C4, no rotation correction was performed since the C2 and C4 locations in the model aligned with the vertical axis relatively well. Since the data was obtained from 2D images, all motions were assumed to be on the mid-sagittal plane.  66  5.2. Inverse Simulation of Swallowing Motion  (a) Fluoroscopy  (b) Model  Figure 5.7: Landmark locations in the VF and model. (a) Stavness et al, 2009 [122], by permission.  Table 5.1: Landmark definitions. Stavness et al, 2009 [122], by permission.  5.2.2  Inverse Controller  An inverse controller was developed as a feature of the ArtiSynth simulation software by Ian Stavness [123]. The function of the controller is to compute muscle activations automatically to track target kinematic trajectories of objects in a model. The formulation of this controller allows inverse simulation on a model with coupled rigid and soft-tissue (FEM) structures. To 67  5.2. Inverse Simulation of Swallowing Motion configure the inverse controller, the user must specify the following: 1. A list of target objects to be tracked 2. A list of muscle exciters 3. The target trajectory for each target object 4. Weight parameters for inverse computation Various components of the model could serve as a target object for tracking: rigid bodies, nodes of FEM mesh, and markers. While the controller is capable of tracking multiple targets simultaneously, the HA marker attached on the surface of hyoid (rigid body) was the only target object used in this study. The controller computes the activations for the list of muscle exciters specified by the user. A muscle exciter is a collection of excitable components in the model. This collection could be made of a single muscle fiber, muscle bundles, or even multiple FEM meshes. Any excitable components not included in this list would not be controlled by the inverse controller and thus would act passively in the simulation. Two weight parameters were used to control the trade-off between minimizing tracking error and minimizing activations. Appropriate values of weight parameters could be model or problem dependent and should be adjusted by the user to suit the purpose of simulation.  5.2.3  Methodology  There were two parts to this experiment and the following simulation settings were shared. First, each of the muscle pairs was grouped as one single exciter to be controlled by the inverse controller. Since only 2D measurements could be obtained from VF, all movements off the mid-sagittal plane were assumed to be negligible. Grouping of muscles on both sides ensured that the motion of the model satisfies this assumption. It also simplified the inverse problem and thus shortened the simulation time. The weight  68  5.2. Inverse Simulation of Swallowing Motion parameters of the inverse controller along with the basic simulation configurations, including the maximum time step and gravity, were kept identical throughout the trials. Experiment 1 This experiment tested the model’s ability to produce realistic hyoid motion for a normal swallow cycle. Hyoid (HA) trajectories acquired from VF were used as target trajectories to be tracked by the inverse controller. A simulation trial was performed on each trajectory data obtained from the three subjects. Table 5.2 summarizes the trials of this experiment. All supra- and infrahyoid muscles, except the thyrohyoid muscle, were utilized by the inverse controller. Thyrohyoid approximation is an important action in swallowing. However, the focus of this study is on the motion of hyoid alone, and therefore, the thyrohyoid muscle was excluded as an exciter for the inverse controller to avoid complications in the simulation. Trial Target data  Muscle exciters  1 2 3 Subject 1 Subject 2 Subject 3 Digastric anterior, Digastric posterior, Geniohyoid, Mylohyoid, Omohyoid inferior, Omohyoid superior, Sternohyoid, Sternothyroid, Stylohyoid  Table 5.2: Inverse simulation trials for experiment 1.  Experiment 2 The goal of the second experiment was to study selective usage of the hyoid muscles. In contrast to experiment 1, the same trajectory data (from Subject  69  5.2. Inverse Simulation of Swallowing Motion Trial Target data  Muscle exciters  1 Digastric anterior Digastric posterior Geniohyoid Mylohyoid Stylohyoid  √ √ √ √ √  2 √ √ √ √  3 4 5 Subject 1 √ √ √ √ √ √ √ √ √ √ √ √  6 √ √ √ √  7  √ √  Table 5.3: Inverse simulation trials for experiment 2.  1) was used by all simulation trials in experiment 2. Instead, the muscle exciters controlled by the inverse controller varied across trials. The trails are summarized in Table 5.3. Subject 1 was selected because its trajectory best resembles the triangular motion commonly described in the literature (see Chapter 2.2). Note that only suprahyoid muscles were considered in this experiment since experiment 1 showed little to no activation of the infrahyoid muscles in all three trials (see Section 5.2.4). The first trial incorporated all suprahyoid muscles to provide a baseline for comparison. Trials 2–6 each excluded one of the five suprahyoid muscle pairs so the significance of each muscle during hyoid excursion could be estimated. In trial 7, geniohyoid and stylohyoid muscles represented muscles that move the hyoid anteriorly and superiorly, respectively. This trial was intended to demonstrate whether the hyoid was capable of achieving excursion with a minimal set of muscles.  5.2.4  Results  Experiment 1 Displacements of the HA marker in the model were recorded and compared against the target data. Figure 5.8 shows the hyoid trajectories for each of the three trails. The blue dashed line represents the target trajectory, whereas the red solid line represents the trajectory produced by the model. The duration of simulations was 3 s. The green solid circle marks the initial position (at 0 s) of the HA marker, whereas the green square indicates the final position (at 3 s). As the plots indicate, the model was able to follow the 70  5.2. Inverse Simulation of Swallowing Motion target trajectory relatively well when the displacement was within a certain range. Once the trajectory exceeded this range, the marker failed to follow the target motion. Figure 5.10 demonstrates that as the target traveled further away from its rest position, the model’s ability to follow the target trajectory diminished. A mean error of 1.89 mm with a standard deviation of 3.12 mm was found. As the figure indicates, the error was either close to or below the mean error for most of the time steps. A maximum error of 2.5 mm was maintained as long as the target trajectory was within a range of 15 mm from the rest position, which translates into an effective range of 10.6 mm in the superior and inferior directions. This is also the region where the target spent most of the time during a swallowing cycle according to the plot. The corresponding muscle activations computed by the inverse controller are plotted in Figure 5.9. The displacements in the anterior and superior directions are shown by the first two columns in blue. The columns in red represent the muscle excitations, which are proportional to the maximum active force of the muscles. The excitation has a range of 0–1, where 0 corresponds to no activation with zero active force and 1 corresponds to full activation with the maximum active force. Only the suprahyoid muscle results are drawn in the figure since the computed activations for the infrahyoid muscles were either zero or insignificantly small. The computed muscle activations showed similar patterns across subjects, considering the variations in the target trajectories. As an example, Figure 5.11 shows the motion of the model from the mid-sagittal view for one of the trials.  71  5.2. Inverse Simulation of Swallowing Motion  (a) Trial 1  (b) Trial 2  (c) Trial 3  Figure 5.8: Hyoid trajectories from experiment 1. The blue dashed line represents the target trajectory, whereas the red solid line represents the trajectory produced by the model. 72  5.2. Inverse Simulation of Swallowing Motion  (a) Trial 1  (b) Trial 2  (c) Trial 3  Figure 5.9: Muscle activations from experiment 1. Blue plots are the hyoid displacements and the red plots are the activation levels of each suprahyoid muscle. 73  5.2. Inverse Simulation of Swallowing Motion  Figure 5.10: This graph shows the error of the motion, which was computed as the Euclidean distance between the target position and the resulting HA marker position, with respect to the Euclidean distance of the target from its rest position. This demonstrates that as the target traveled further, the model’s ability to follow its trajectory was reduced. Note that each point in the plot represents a 0.01 s time step in the simulation. µ = mean error and σ = standard deviation.  74  5.2. Inverse Simulation of Swallowing Motion  (a) t=0.6sec  (b) t=0.8sec  (c) t=1.0sec  (d) t=1.2sec  (e) t=1.4sec  (f) t=1.6sec  (g) t=1.8sec  (h) t=2.0sec  (i) t=2.2sec  (j) t=2.4sec  (k) t=2.6sec  (l) t=2.8sec  (m) t=3.0sec  Figure 5.11: Demonstration of the model in motion during inverse simulation of Trial 1. Blue line represents the target data and green line represents tracking of the HA marker. 75  5.2. Inverse Simulation of Swallowing Motion Experiment 2 The resulting displacements for all trials along with the target trajectory are plotted together in Figure 5.12. Minor differences could be observed across trails except for trial 4, where the geniohyoid muscle was excluded by the inverse controller. The considerable decrease in the anterior displacement in trial 4 could signify the importance of the geniohyoid muscle in propelling the hyoid forward. Muscle activations corresponding to each of the trials are shown in Figure 5.13. As the results indicate, there was an exchange of workload between muscles within the same function group across trials. Muscles responsible for hyoid elevation (digastric posterior, mylohyoid, and stylohyoid) compensated for each other when one of them was excluded by the controller. There was also an increase in activation of the digastric anterior muscle when the geniohyoid muscle was excluded. These two muscles were responsible for moving the hyoid anteriorly.  Figure 5.12: Hyoid trajectories from experiment 2.  76  5.2. Inverse Simulation of Swallowing Motion  (a) Trial 1  (b) Trial 2  (c) Trial 3  (d) Trial 4  (e) Trial 5  (f) Trial 6  (g) Trial 7  Figure 5.13: Muscle activations from experiment 2.  77  5.2. Inverse Simulation of Swallowing Motion  5.2.5  Discussion  As the results demonstrate, the model was able to achieve hyoid excursion that closely resembles real hyoid movements measured from healthy subjects. While there was a range within which the model performed best, the overall hyoid trajectory produced maintains a shape similar to that of the target data. There are several possible explanations for the limited range of accurate hyoid motion. First, there were morphological differences between the model and the subjects that could have an effect on the range of hyoid motion. For instance, if the distance between the hyoid and mandible of the model was much shorter than of the subjects, there would be less space for the hyoid to move and shorter contraction distance could be made by the muscles. As an example, Figure 5.14 shows a lateral view of the model and one of the VF subjects side by side. With C2–C4 as the reference axis, the hyoid position in the VF image is lower relative to the bottom of the mandible as compared to the hyoid in the model. Since hyoid excursion itself commonly has a range of motion of around 10–20 mm, even small differences such as that shown in the images could have an impact on the motion. In addition, these morphological differences could be a factor in determining how far the hyoid has to travel to successfully complete pharyngeal swallowing. In addition, the rest position of the model was assumed to be the position of the male cadaver in the VHP images which could be different from the rest positions of the VF subjects. Consequently, the relative position of the hyoid to the surrounding structures as well as the initial length of muscles would likely differ. Since the motion of the hyoid is greatly constrained by its coupling with the surrounding structures, differences in posture could in part explain the range limitation. Nevertheless, hyoid motion range has been observed to be highly variable, as was discussed in Section 2.2. The 10.6 mm range of displacement in both anterior and superior directions offered by the model is an acceptable range that is not uncommon in healthy subjects studied by past researchers (see Table C.1 for a list of hyoid displacement reported by past studies). According to the simulation results, the geniohyoid, mylohyoid, and sty-  78  5.2. Inverse Simulation of Swallowing Motion  (a)  (b)  Figure 5.14: Example of morphological difference between individuals. (a) Fluoroscopic image of Subject 2 by [122], (b) the model with geometries from VHP.  lohyoid muscles seemed to be the major contributors to hyoid maneuvers when all muscles could be used. However, the system offers a large number of degrees of freedom, where only the motion of the hyoid has to be satisfied. Hence, the set of possible muscle activities is also large. It is only natural that there is large intersubject variability in the muscle activations observed in various studies [47, 120]. The muscle activations computed by the inverse controller were only one of the potential solutions for producing the result trajectory, which in our experiment, was relatively invariant across the target data from different subjects. However, the computed muscle activations could change dramatically when parameters of the inverse controller were adjusted. In experiment 2, the inverse simulations confirmed that subsets of the muscles were sufficient for achieve hyoid excursion. In fact, most of the displacement could be produced by the geniohyoid and stylohyoid muscles alone. Our finding agrees with that of [120]. In their EMG study, the geniohyoid, mylohyoid, and digastric anterior muscle were used selectively by the subjects during swallowing. The model again demonstrated its ability to produce realistic swallowing behavior that preserved the characteristics 79  5.2. Inverse Simulation of Swallowing Motion of the anatomical system. The result also suggested that it is possible to minimize swallowing difficulty caused by dysfunction of some of the suprahyoid muscles with physical therapy. Such dysfunction could result from physical damages to the muscle tissues or nerves. The patients could be trained to better use or strengthen the remaining functional muscles to compensate for the damaged ones.  80  Chapter 6  Evaluation II: Hyoid Motion and Muscle Aging The ability of the model to produce realistic hyoid excursion during swallowing was verified in the last chapter. In this chapter, the model is further utilized to study the effect of muscle strength loss on hyoid range of motion. Muscles responsible for elevating the hyoid bone and moving it toward the mandible were activated under different force settings to simulate the maximum displacement of the hyoid. The reduction in hyoid motion range was then estimated for individuals of various age based on the muscle strength loss associated with each ages group.  6.1  Hypothesis  The objective of this study was to examine the effects of muscle aging on the mobility of the hyoid bone using our model. There exist many age-related physiological changes that would likely have an impact on the biomechanics of the hyolaryngeal complex. This study focused on one particular aspect of the aging process, loss of muscle strength, since the performance of actuators in a mechanical system such as the hyolaryngeal complex would likely have most impact on the system’s behavior. The experiment was formulated under the following hypothesis: The loss of strength in the suprahyoid muscles due to the aging process directly impacts the range of hyoid motion in the anterior and superior directions.  81  6.2. Methodology Note that depression of the hyoid from its resting position was not studied since such motion is mainly observed during singing but not swallowing.  6.2  Methodology  As in all other experiments described in this thesis, the simulations performed for this section utilized the basic settings specified in Chapter 5, where acceleration due to gravity was set to 9.8 m/s2 in the z-direction and the maximum time step was 1.0 ms. Two sets of simulation trials were performed. The first simulated the maximum distance that could be travelled by the hyoid in the anterior direction, and the other simulated the maximum distance in the superior direction. To simulate the maximum displacement, a set of selected muscles was activated with a smoothed step-up function (Figure 6.1) in each trial. Table 6.1 summarizes the two trial sets. For Trial Set 1, the digastric anterior and geniohyoid muscles were activated to find the maximum anterior displacement. The rest of the suprahyoid muscles— digastric posterior, mylohyoid, and stylohyoid—were activated in Trial Set 2 to find the maximum superior displacement. This choice of muscles was made based on the results from Chapter 5. The simulations were designed to run for 1 s so the model would have enough time to reach a steady state. The HA marker, which is described in Chapter 5, was again used for measuring the movement of the hyoid. The position of the HA marker was recorded and used to obtain the distance traveled by the hyoid. In both sets of trials, the maximum displacement was computed by finding the Euclidean distance between the initial and finial HA marker position. The muscles in Table 6.1 were selected so the hyoid would produce motions ideally only in the direction of interest; however, minor displacements in the other directions were expected. The error between the Euclidean distance and the displacement in the direction of interest (horizontal for Trail Set 1 and vertical for Trial Set 2) was computed to provide a measure of how well the motion was completed. Loss of muscle strength due to aging was incorporated in the experiment by adjusting the maximum active force of each muscle in the model. Based 82  6.2. Methodology  Figure 6.1: Step-up function for muscle activation. Trial set Direction of interest Muscles activated  1 Anterior Digastric anterior, Geniohyoid,  2 Superior Digastric posterior, Mylohyoid, Stylohyoid  Table 6.1: Simulation trials for finding the maximum hyoid displacement.  on [74], the maximum force was assumed to have a linear decline rate of 12% per decade from age 50 onward. Before age 50, the maximum force was assumed to be constant as defined in Table B.3. For both trial sets, a series of simulations was repeatedly performed with various muscle force settings. In every iteration, the maximum active force of each muscle was reduced by 10% of its original value. Besides the rate of muscle strength decline, several other assumptions were made in this experiment. To isolate the effect of muscle weakening on hyoid mobility from other factors, the loss of muscle strength is assumed not to be associated with other physiological changes caused by aging, such as tissue property changes or loss of muscle mass. It is also assumed that no morphological changes occur with aging, which is uncommon in reality. Finally, it is assumed that anterior and superior hyoid motions are achieved exclusively by the muscles listed in Table 6.1, and the surrounding structures not included in the model have little to no impact on hyoid mobility. 83  6.3. Results  6.3  Results  The obtained maximum hyoid displacements in both anterior and superior directions are plotted in Figure 6.2. The displacements are expressed in millimeters as well as in percentage. As shown in the graph, the maximum displacement decreased with a decline in maximum muscle force. The rate of this reduction accelerated as the muscle force diminished and the displacement eventually reached zero. When looking the percentage plot, the reduction in maximum displacement along the superior direction occurred slightly faster than the reduction along the anterior direction. Since more and stronger muscles were used to produce the superior motion in this experiment, a difference in rate between the two result curves was expected. Under the assumption of a 12% loss of muscle strength per decade, the results were interpolated to show the relationship between maximum displacement and age, as shown in Figure 6.3. Based on this plot, a person would experience an 20% reduction in hyoid range of motion if they lived to be a 100 years old. Between ages 50 and 100, the respective muscle strengths decreased from 100% to 40%, which accounts for the most linear portion of the curve in 6.2. Error between the Euclidean distance and displacement in the direction of interest was within 0.99% and 1.28% for Trial Sets 1 and 2, respectively.  6.4  Discussion  The results of this experiment support our hypothesis that suprahyoid muscle strength can directly influence the range of hyoid motion. As the muscles become weaker, the ability for the hyoid to move freely would be impacted, and thus introduce risks when swallowing. However, since the simulations showed a range reduction of only around 20% over a lifespan of 100 years, which is somehow smaller than that reported by Kim [60] and Logemann [70], there must exist other factors that further reduce the hyoid mobility of an aged person. Our results also suggest that severe muscle weakening might be required 84  6.4. Discussion  (a)  (b)  Figure 6.2: Maximum hyoid displacement with respect to muscle strength. Displacement is expressed in (a) millimeter and (b) percentage.  85  6.4. Discussion  (a)  (b)  Figure 6.3: Maximum hyoid displacement with respect to age. Displacement is expressed in (a) millimeter and (b) percentage.  86  6.4. Discussion to reduce hyoid excursion sufficiently to impairing swallowing function significantly. Dysphagia patients studied by Kendall [58], Paik [90], and Wang [134] all demonstrated significant reductions in hyoid motion compared to healthy subjects. Mild muscle weakening alone would not be enough to produce such reductions according to this experiment. Perhaps other agerelated physiological changes in muscles, such as atrophy or loss of tissue elasticity, in combination with loss of strength are required to cause significant range reductions. Age-associated morphological changes in the head and neck region as reported by [125] could also be a factor. While the loss of muscle strength might not be the primary reason for age-induced hyoid motion range reduction, it could still lower the ability to swallow in people of very old age. Based on past studies by Pyka et al. [102] and Skelton et al.  [116], resistance training can be carried out  for elderly people as an effective means to increase muscle strength. While resistance training is difficult to perform for the neck muscles, isometric training such as Shaker’s Exercise [109] has been designed to strengthen the suprahyoid muscle and has been shown to improve swallowing quality in elderly individuals with or without dysphagia [27, 72]. This provides a potential solution for minimizing the effect of muscle weakening on hyoid mobility. As discussed in Chapter 2, there is strong evidence suggesting that agerelated physiological changes are some of the main causes behind the increased prevalence of dysphagia among the elderly. The end goal of this experiment was to use the range of hyoid motion as an indicator of dysphagia and to better understand the risk of oropharyngeal dysphagia among various age groups. While this experiment looked at a very specific and narrow aspect of the swallowing and aging process, the results could provide some insight for future studies on estimating how individuals are prone to swallowing disorders.  87  Chapter 7  Conclusion 7.1  Summary  While the ability to swallow is crucial in maintaining adequate nutrition, there is a high prevalence of dysphagia among the elderly, which is associated with a high mortality rate. The goal of this thesis was to attempt to understand the effects of age on oropharyngeal swallowing. As a representative sign of the aging process, this study focused on the loss of muscle strength and its relationship with swallowing ability. The specific approach was to use hyoid mobility, specifically the range of motion, as an indicator for evaluating the ability to perform normal swallows. A 3D biomechanical model of the hyolaryngeal complex was developed based on anatomical data of a male cadaver. The model was validated against current knowledge of the extrinsic laryngeal muscle functions. The model’s ability to produce realistic hyoid motion for swallowing was then demonstrated with inverse simulation. Finally, the effect of muscle weakening on the range of hyoid motion was examined using the model. The contributions of this thesis are summarized below: 1. Creation of a 3D biomechanical model of the hyolaryngeal complex. Based on geometries extracted from the Visible Human Project dataset and the use of the ArtiSynth simulation platform, an anatomically accurate computer model of the larynx coupled with extrinsic laryngeal muscles was created. Functions of the muscles in the model were examined and were found to be consistent with current knowledge. The model is a state-of-the-art tool for studying the biomechanics of hyoid and laryngeal tasks such as swallowing and speech 88  7.2. Future Directions production. 2. Identification of a workflow for computer model generation. A procedure of creating a physically based dynamic model of a generic musculoskeletal system from medical images was developed. The workflow is specific to the simulation platform ArtiSynth, and incorporates the use of rigid body, spring, and finite-element models. Techniques that assist in the process of model making, such as symmetry making and volumetric mesh fitting, were employed. 3. Inverse simulation of hyoid excursion during normal swallowing. Clinically measured data from videofluoroscopy were used to inversely compute possible muscle activities for producing desired hyoid motion using the hyolaryngeal model. The model was capable of mimicking realistic hyoid motion within a certain range. Selective usage of the suprahyoid was demonstrated to be possible and sufficient in achieving normal hyoid excursion. 4. Study of the effects of age-related muscle weakening on oropharyngeal swallowing. The range of hyoid motion in the anterior and superior direction under various muscle strengths was simulated. The range was found to decrease with the muscle strength. The relationship between the range of motion and muscle strength was curvilinear, with a gradually increasing rate of decline. Assuming a life expectancy of 100 years, the model computed a range reduction of around 20% due to muscle weakening alone.  7.2  Future Directions  There are many improvements and extensions that are possible with this work. Here lists some promising directions for future investigations. Improve the way muscle aging was modeled. Muscle weakening was the only age-related factor considered in this thesis and it was modeled as a simple linear decline of strength. However, muscle aging is a much more 89  7.2. Future Directions complicated process that can result in a number of physiological changes in the human body, including changes in muscle material properties. If the modeling of this process is perfected, its effect on oropharyngeal swallowing can be better understood. Perform clinical validation of the model. While the model was examined using current knowledge of the head and neck physiology, quantitative evaluation against clinical data is still necessary to ensure the model is valid. This is especially true if the model is to be used in any medical application such as treatment planning, where extensive validation would provide a measure of accuracy on the model’s predications. A combination of EMG and 3D dynamic imaging techniques such as MRI or CT would likely be required to fully validate the model. Obtain accurate material property data for the tissues. Many of the tissue property parameters utilized by the model were approximations made based on educated guesses. The correctness of a computer model relies heavily on the parameters that characterize the system. As of today, mechanical analysis on tissue specimens is the most accurate, if not the only, way for measuring most of the material properties. However, analysis of this type has rarely been performed on tissues of the head and neck region. Extend the model to incorporate other structures. Some passive structures such as the thyroid glands, adipose tissues, and mucous membrane were excluded from our model for simplicity. However, they might have minor influences on the biomechanics of the system. Furthermore, the hyolaryngeal complex is highly coupled with the surrounding structures, including the tongue and pharynx. Integration of the current model with models of other upper airway components would allow simulation of the entire swallowing process. Many research questions regarding swallowing can then be answered. For example, the mechanism of epiglottic closure has been under debate for the past few decades and could be elucidated with such a model. Develop patient-specific modeling for the purpose of treatment planning. A potential application for the biomechanical model is as a tool for treatment planning. The model can be used to predict surgical effects 90  7.2. Future Directions on swallowing ability after head and neck operations. To do so, a process for patient-specific adaptation of the model needs to be developed.  91  Bibliography [1] K.F. Andersen and A. Sonninen. The function of the extrinsic laryngeal muscles at different pitch. Acta Oto-Laryngologica, 51(1-2):89–93, 1960. 7→ page 13 [2] GM Ardran and FH Kemp. The mechanism of the larynxii the epiglottis and closure of the larynx. British Journal of Radiology, 40(473):372–389, 1967. 7→ page 10 [3] B.M. Auerbach and C.B. Ruff. Limb bone bilateral asymmetry: variability and commonality among modern humans. Journal of Human Evolution, 50(2):203–218, 2006. 7→ page 33 [4] İ. Aydoğdu, N. Yüceyar, N. Kiylioğlu, S. Tarlaci, Y. Seçil, M. Pehelivan, and C. Ertekin. Physiological changes in oropharyngeal swallowing with age: An electrophysiological study. Journal of Neurological Sciences (Turkish), 24(2):144–154, 2007. 7→ page 14 [5] J.M. Barkmeier, S. Bielamowicz, N. Takeda, and C.L. Ludlow. Laryngeal activity during upright vs. supine swallowing. Journal of Applied Physiology, 93(2):740–745, 2002. 7→ page 13 [6] S.E. Benzley, E. Perry, K. Merkley, B. Clark, and G. Sjaardama. A comparison of all hexagonal and all tetrahedral finite element meshes for elastic and elasto-plastic analysis. In Proceedings, 4th International Meshing Roundtable, volume 17, pages 179–191. Citeseer, 1995. 7→ page 41 [7] T. Blacker. Meeting the challenge for automated conformal hexahedral meshing. In 9th International Meshing Roundtable, pages 11–20. Citeseer, 2000. 7→ page 20 [8] S.S. Blemker and S.L. Delp. Three-dimensional representation of complex muscle architectures and geometries. Annals of biomedical engineering, 33(5):661–673, 2005. 7→ page 41  92  Bibliography [9] S.S. Blemker, P.M. Pinsky, and S.L. Delp. A 3d model of muscle reveals the causes of nonuniform strains in the biceps brachii. Journal of biomechanics, 38(4):657–665, 2005. 7→ page 55 [10] T. Breyer, M. Echternach, S. Arndt, B. Richter, O. Speck, M. Schumacher, and M. Markl. Dynamic magnetic resonance imaging of swallowing and laryngeal motion using parallel imaging at 3 t. Magnetic resonance imaging, 27(1):48–54, 2009. 7→ page 16 [11] S. Buchaillard, P. Perrier, and Y. Payan. A biomechanical model of cardinal vowel production: Muscle activations and the impact of gravity on tongue positioning. The Journal of the Acoustical Society of America, 126:2033, 2009. 7→ page 25, 117 [12] M. Bucki, C. Lobos, and Y. Payan. A fast and robust patient specific finite element mesh registration technique: application to 60 clinical cases. Medical Image Analysis, 14(3):303–317, 2010. 7→ page 30, 34 [13] WE Burke, WW Tuttle, CW Thompson, CD Janney, and RJ Weber. The relation of grip strength and grip-strength endurance to age. Journal of applied physiology, 5(10):628–630, 1953. 7→ page 15 [14] T.A. Burnett, E.A. Mann, S.A. Cornell, and C.L. Ludlow. Laryngeal elevation achieved by neuromuscular stimulation at rest. Journal of Applied Physiology, 94(1):128–134, 2003. 7→ page 12 [15] MJ Campbell, AJ McComas, and F. Petito. Physiological changes in ageing muscles. Journal of Neurology, Neurosurgery & Psychiatry, 36(2):174–182, 1973. 7→ page 15 [16] M.W. Chang, E. Lin, and J.N. Hwang. Contour tracking using a knowledge-based snake algorithm to construct three-dimensional pharyngeal bolus movement. Dysphagia, 14(4):219–227, 1999. 7→ page 17 [17] IJ Cook, WJ Dodds, RO Dantas, B. Massey, MK Kern, IM Lang, JG Brasseur, and WJ Hogan. Opening mechanisms of the human upper esophageal sphincter. American Journal of PhysiologyGastrointestinal and Liver Physiology, 257(5):G748–G759, 1989. 7→ page 10 [18] I.J. Cook and P.J. Kahrilas. Aga technical review on management of oropharyngeal dysphagia. Gastroenterology, 116(2):455–478, 1999. 7→ page 1, 2 93  Bibliography [19] IJ Cook, MD Weltman, K. Wallace, DW Shaw, E. McKay, RC Smart, and SP Butler. Influence of aging on oral-pharyngeal bolus transit and clearance during swallowing: scintigraphic study. American Journal of Physiology-Gastrointestinal and Liver Physiology, 266(6):G972–G977, 1994. 7→ page 13, 14 [20] J.E. Croghan, E.M. Burke, S. Caplan, and S. Denman. Pilot study of 12-month outcomes of nursing home patients with aspiration on videofluoroscopy. Dysphagia, 9(3):141–146, 1994. 7→ page 1 [21] T. C̀uk, P. Leben-Seljak, and M. Stefaneie. Lateral asymmetry of human long bones. Variability and Evolution, 9:19–32, 2001. 7→ page 33 [22] A. Daggett, J. Logemann, A. Rademaker, and B. Pauloski. Laryngeal penetration during deglutition in normal subjects of various ages. Dysphagia, 21(4):270–274, 2006. 7→ page 13 [23] R.O. Dantas, M.K. Kern, B.T. Massey, WJ Dodds, PJ Kahrilas, JG Brasseur, IJ Cook, and IM Lang. Effect of swallowed bolus variables on oral and pharyngeal phases of swallowing. American Journal of Physiology-Gastrointestinal and Liver Physiology, 258(5):G675– G681, 1990. 7→ page 11, 121 [24] MP De Vries, HK Schutte, and GJ Verkerke. Determination of parameters for lumped parameter models of the vocal folds using a finiteelement method approach. The Journal of the Acoustical Society of America, 106:3620, 1999. 7→ page 16 [25] WJ Dodds, KM Man, IJ Cook, PJ Kahrilas, ET Stewart, and MK Kern. Influence of bolus volume on swallow-induced hyoid movement in normal subjects. American Journal of Roentgenology, 150(6):1307–1309, 1988. 7→ page 10, 11, 121 [26] T.J. Doherty, A.A. Vandervoort, and W.F. Brown. Effects of ageing on the motor unit: a brief review. Canadian journal of applied physiology, 18(4):331–358, 1993. 7→ page 14, 15 [27] C. Easterling, B. Grande, M. Kern, K. Sears, and R. Shaker. Attaining and maintaining isometric and isokinetic goals of the shaker exercise. Dysphagia, 20(2):133–138, 2005. 7→ page 87  94  Bibliography [28] HE Eckel, C. Sittel, P. Zorowka, and A. Jerke. Dimensions of the laryngeal framework in adults. Surgical and Radiologic Anatomy, 16(1):31– 36, 1994. 7→ page 16, 33 [29] A. Eisbruch, T. Lyden, C.R. Bradford, L.A. Dawson, M.J. Haxer, A.E. Miller, T.N. Teknos, D.B. Chepeha, N.D. Hogikyan, J.E. Terrell, et al. Objective assessment of swallowing dysfunction and aspiration after radiation concurrent with chemotherapy for head-and-neck cancer. International Journal of Radiation Oncology* Biology* Physics, 53(1):23–28, 2002. 7→ page 8 [30] O. Ekberg and M.J. Feinberg. Altered swallowing function in elderly patients without dysphagia: radiologic findings in 56 cases. American Journal of Roentgenology, 156(6):1181–1184, 1991. 7→ page 4, 13 [31] O. Ekberg, S. Hamdy, V. Woisard, A. Wuttge-Hannig, and P. Ortega. Social and psychological burden of dysphagia: its impact on diagnosis and treatment. Dysphagia, 17(2):139–146, 2002. 7→ page 1 [32] O. Ekberg and S.V. Sigurjónsson. Movement of the epiglottis during deglutition. Abdominal Imaging, 7(1):101–107, 1982. 7→ page 10 [33] W.J. Evans and J. Lexell. Human aging, muscle mass, and fiber type composition. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 50(Special Issue):11, 1995. 7→ page 15 [34] S. Fels, F. Vogt, K. Van Den Doel, J. Lloyd, I. Stavness, and E. Vatikiotis-Bateson. Artisynth: A biomechanical simulation platform for the vocal tract and upper airway. In International Seminar on Speech Production, Ubatuba, Brazil. Citeseer, 2006. 7→ page 30 [35] B.R. Fink, R.W. Martin, and C.A. Rohrmann. Biomechanics of the human epiglottis. Acta Oto-Laryngologica, 87(3-6):554–559, 1979. 7→ page 10 [36] W.R. Frontera, V.A. Hughes, R.A. Fielding, M.A. Fiatarone, W.J. Evans, and R. Roubenoff. Aging of skeletal muscle: a 12-yr longitudinal study. Journal of Applied Physiology, 88(4):1321–1326, 2000. 7→ page 15 [37] N. Fujii, Y. Inamoto, E. Saitoh, M. Baba, S. Okada, S. Yoshioka, T. Nakai, Y. Ida, K. Katada, and J.B. Palmer. Evaluation of swallowing using 320-detector-row multislice ct. part i: Single-and multiphase 95  Bibliography volume scanning for three-dimensional morphological and kinematic analysis. Dysphagia, 26(2):99–107, 2011. 7→ page 16 [38] A. Gasiorowska and R. Fass. Current approach to dysphagia. Gastroenterology & Hepatology, 5(4):269, 2009. 7→ page 1 [39] B.H. Goodpaster, S.W. Park, T.B. Harris, S.B. Kritchevsky, M. Nevitt, A.V. Schwartz, E.M. Simonsick, F.A. Tylavsky, M. Visser, and A.B. Newman. The loss of skeletal muscle strength, mass, and quality in older adults: the health, aging and body composition study. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 61(10):1059–1064, 2006. 7→ page 14, 15, 16 [40] H.E. Gunter. A mechanical model of vocal-fold collision with high spatial and temporal resolution. The Journal of the Acoustical Society of America, 113:994, 2003. 7→ page 16 [41] K.E. Hall. Ii. neural control of the aging gut: can an old dog learn new tricks? American Journal of Physiology-Gastrointestinal and Liver Physiology, 283(4):G827–G832, 2002. 7→ page 4 [42] A.G. Hannam, I. Stavness, J.E. Lloyd, and S. Fels. A dynamic model of jaw and hyoid biomechanics during chewing. Journal of biomechanics, 41(5):1069–1076, 2008. 7→ page 17, 118 [43] P. Harper, S.S. Kraman, H. Pasterkamp, and G.R. Wodicka. An acoustic model of the respiratory tract. Biomedical Engineering, IEEE Transactions on, 48(5):543–550, 2001. 7→ page 116, 118 [44] AV Hill. The heat of shortening and the dynamic constants of muscle. Proceedings of the Royal Society of London. Series B, Biological Sciences, 126(843):136–195, 1938. 7→ page 57 [45] K. Ho-Le. Finite element mesh generation methods: a review and classification. Computer-aided design, 20(1):27–38, 1988. 7→ page 20 [46] Y. Honda and N. Hata. Dynamic imaging of swallowing in a seated position using open-configuration mri. Journal of Magnetic Resonance Imaging, 26(1):172–176, 2007. 7→ page 16 [47] A.W. Hrycyshyn and J.V. Basmajian. Electromyography of the oral stage of swallowing in man. American Journal of Anatomy, 133(3):333–340, 1972. 7→ page 13, 79 96  Bibliography [48] A. Hu, T. Wilson, H. Ladak, P. Haase, and K. Fung. Threedimensional educational computer model of the larynx: voicing a new direction. Archives of Otolaryngology- Head and Neck Surgery, 135(7):677, 2009. 7→ page 17 [49] T.J. Hughes. The finite element method: Linear static and dynamic finite element analysis. Recherche, 67:02, 2000. 7→ page 40 [50] E.J. Hunter and I.R. Titze. Individual subject laryngeal dimensions of multiple mammalian species for biomechanical models. The Annals of otology, rhinology & laryngology, 114(10):809–818, 2005. 7→ page 16 [51] E.J. Hunter, I.R. Titze, and F. Alipour. A three-dimensional model of vocal fold abduction/adduction. The Journal of the Acoustical Society of America, 115:1747, 2004. 7→ page 16 [52] Y. Inamoto, N. Fujii, E. Saitoh, M. Baba, S. Okada, K. Katada, Y. Ozeki, D. Kanamori, and J.B. Palmer. Evaluation of swallowing using 320-detector-row multislice ct. part ii: Kinematic analysis of laryngeal closure during normal swallowing. Dysphagia, pages 1–9, 2010. 7→ page 16 [53] R. Ishida, J.B. Palmer, and K.M. Hiiemae. Hyoid motion during swallowing: factors affecting forward and upward displacement. Dysphagia, 17(4):262–272, 2002. 7→ page 11, 121 [54] P. Jacob, PJ Kahrilas, JA Logemann, V. Shah, T. Ha, et al. Upper esophageal sphincter opening and modulation during swallowing. Gastroenterology, 97(6):1469, 1989. 7→ page 10 [55] P.J. Kahrilas, S. Lin, J. Chen, and J.A. Logemann. Three-dimensional modeling of the oropharynx during swallowing. Radiology, 194(2):575– 579, 1995. 7→ page 17 [56] D.A. Kallman, C.C. Plato, and J.D. Tobin. The role of muscle loss in the age-related decline of grip strength: cross-sectional and longitudinal perspectives. Journal of Gerontology, 45(3):M82–M88, 1990. 7→ page 15 [57] P.M. Kellen, D. Becker, J.M. Reinhardt, and D. van Daele. Tracking the hyoid bone in videofluoroscopic swallowing studies. In Proceedings of SPIE, the International Society for Optical Engineering, pages 97  Bibliography 69142D–1. Society of Photo-Optical Instrumentation Engineers, 2008. 7→ page 16 [58] K.A. Kendall and R.J. Leonard. Hyoid movement during swallowing in older patients with dysphagia. Archives of Otolaryngology- Head and Neck Surgery, 127(10):1224, 2001. 7→ page 87, 121 [59] M. Kern, E. Bardan, R. Arndorfer, C. Hofmann, J. Ren, R. Shaker, et al. Comparison of upper esophageal sphincter opening in healthy asymptomatic young and elderly volunteers. The Annals of otology, rhinology, and laryngology, 108(10):982, 1999. 7→ page 14 [60] Y. Kim and G.H. McCullough. Maximum hyoid displacement in normal swallowing. Dysphagia, 23(3):274–279, 2008. 7→ page 3, 10, 11, 84, 121 [61] A. Kirzinger and U. Jürgens. Role of extralaryngeal muscles in phonation of subhuman primates. Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral Physiology, 175(2):215–222, 1994. 7→ page 13 [62] M. Kob and C. Butenweg. A finite element model of the interaction between intra-and extralaryngeal muscles. In International Conference on Voice Physiology and Biomechanics, 2004. 7→ page 17 [63] A. Kurosu and J.A. Logemann. Gender effects on airway closure in normal subjects. Dysphagia, 25(4):284–290, 2010. 7→ page 14 [64] GEJ Langenbach and AG Hannam. The role of passive muscle tensions in a three-dimensional dynamic model of the human jaw. Archives of oral biology, 44(7):557–573, 1999. 7→ page 117 [65] L. Larsson, G. Grimby, and J. Karlsson. Muscle strength and speed of movement in relation to age and muscle morphology. Journal of Applied Physiology, 46(3):451–456, 1979. 7→ page 15 [66] R.J. Leonard, K.A. Kendall, S. McKenzie, M.I. Gonçalves, and A. Walker. Structural displacements in normal swallowing: a videofluoroscopic study. Dysphagia, 15(3):146–152, 2000. 7→ page 121 [67] J. Lexell, C.C. Taylor, M. Sjöström, et al. What is the cause of the ageing atrophy? total number, size and proportion of different fiber types studied in whole vastus lateralis muscle from 15-to 83-year-old 98  Bibliography men. Journal of the neurological sciences, 84(2-3):275, 1988. 7→ page 14, 15 [68] RS Lindle, EJ Metter, NA Lynch, JL Fleg, JL Fozard, J. Tobin, TA Roy, and BF Hurley. Age and gender comparisons of muscle strength in 654 women and men aged 20–93 yr. Journal of Applied Physiology, 83(5):1581–1587, 1997. 7→ page 15 [69] J.A. Logemann, P.J. Kahrilas, J. Cheng, BR Pauloski, P.J. Gibbons, A.W. Rademaker, and S. Lin. Closure mechanisms of laryngeal vestibule during swallow. American Journal of PhysiologyGastrointestinal and Liver Physiology, 262(2):G338–G344, 1992. 7→ page 10 [70] J.A. Logemann, B.R. Pauloski, A.W. Rademaker, L.A. Colangelo, P.J. Kahrilas, and C.H. Smith. Temporal and biomechanical characteristics of oropharyngeal swallow in younger and older men. Journal of Speech, Language, and Hearing Research, 43(5):1264, 2000. 7→ page 10, 14, 84, 121 [71] J.A. Logemann, B.R. Pauloski, A.W. Rademaker, and P.J. Kahrilas. Oropharyngeal swallow in younger and older women: videofluoroscopic analysis. Journal of speech, language, and hearing research, 45(3):434, 2002. 7→ page 10, 14, 16, 121 [72] J.A. Logemann, A. Rademaker, B.R. Pauloski, A. Kelly, C. StanglMcBreen, J. Antinoja, B. Grande, J. Farquharson, M. Kern, C. Easterling, et al. A randomized study comparing the shaker exercise with traditional therapy: a preliminary study. Dysphagia, 24(4):403–411, 2009. 7→ page 87 [73] D.S. Lundy, C. Smith, L. Colangelo, P.A. Sullivan, J.A. Logemann, C.L. Lazarus, L.A. Newman, T. Murry, L. Lombard, and J. Gaziano. Aspiration: cause and implications. Otolaryngology-Head and Neck Surgery, 120(4):474–478, 1999. 7→ page 1, 3, 8, 14 [74] N.A. Lynch, EJ Metter, RS Lindle, JL Fozard, JD Tobin, TA Roy, JL Fleg, and BF Hurley. Muscle quality. i. age-associated differences between arm and leg muscle groups. Journal of Applied Physiology, 86(1):188–194, 1999. 7→ page 15, 16, 83 [75] F. Magendie. An elementary compendium of physiology.(translated by e. milligan.) edinburgh: J, 1831. 7→ page 10 99  Bibliography [76] A. Manduca, TE Oliphant, MA Dresner, JL Mahowald, SA Kruse, E. Amromin, JP Felmlee, JF Greenleaf, and RL Ehman. Magnetic resonance elastography: non-invasive mapping of tissue elasticity. Medical Image Analysis, 5(4):237–254, 2001. 7→ page 40 [77] P.E. Marik and D. Kaplan. Aspiration pneumonia and dysphagia in the elderly*. Chest, 124(1):328–336, 2003. 7→ page 1 [78] B. Martin-Harris, M.B. Brodsky, Y. Michel, D.O. Castell, M. Schleicher, J. Sandidge, R. Maxwell, and J. Blair. Mbs measurement tool for swallow impairmentmbsimp: establishing a standard. Dysphagia, 23(4):392–405, 2008. 7→ page 3, 8 [79] K.A. Mays, J.B. Palmer, and K.V. Kuhlemeier. Influence of craniofacial morphology on hyoid movement: a preliminary correlational study. Dysphagia, 24(1):71–76, 2009. 7→ page 10, 11, 12, 121 [80] E.J. Metter, R. Conwit, J. Tobin, and J.L. Fozard. Age-associated loss of power and strength in the upper extremities in women and men. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 52(5):B267, 1997. 7→ page 15 [81] K.W.P. Miller, P.L. Walker, and R.L. O’Halloran. Age and sex-related variation in hyoid bone morphology. Journal of forensic sciences, 43:1138–1143, 1998. 7→ page 5 [82] H. Mizunuma, M. Sonomura, K. Shimokasa, H. Ogoshi, S. Nakamura, and N. Tayama. Numerical modeling and simulation on the swallowing of jelly. Journal of Texture Studies, 40(4):406–426, 2009. 7→ page 16 [83] S.R. Moisik, J.H. Esling, J. Trouvain, and WJ Barry. 3-d auditoryarticulatory modeling of the laryngeal constrictor mechanism. In Proceedings of the 16th International Congress of Phonetic Sciences, volume 1, pages 373–376. Universität des Saarlandes Saarbrücken, 2007. 7→ page 17 [84] S.M. Molfenter and C.M. Steele. Physiological variability in the deglutition literature: hyoid and laryngeal kinematics. Dysphagia, 26(1):67– 74, 2011. 7→ page 10 [85] K.P. Mulcahy, P.C. Langdon, and F. Mastaglia. Dysphagia in inflammatory myopathy: Self-report, incidence, and prevalence. Dysphagia, pages 1–6, 2011. 7→ page 5 100  Bibliography [86] M. Nesme, M. Marchal, E. Promayon, M. Chabanas, Y. Payan, and F. Faure. Physcally realistic interactive simulation for biological soft tissues. Recent Res. Devel. Biomechanics, 2, 2005. 7→ page 41 [87] J.M. Novak, OE Rhodes, M.H. Smith, and R.K. Chesser. Morphological asymmetry in mammals: genetics and homeostasis reconsidered. Acta Theriologica, 38:7–7, 1993. 7→ page 33 [88] CWJ Oomens, M. Maenhout, CH Van Oijen, MR Drost, and FP Baaijens. Finite element modelling of contracting skeletal muscle. Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences, 358(1437):1453–1460, 2003. 7→ page 40 [89] S.J. Owen and S. Saigal. H-morph: an indirect approach to advancing front hex meshing. International Journal for Numerical Methods in Engineering, 49(1-2):289–312, 2000. 7→ page 40 [90] N.J. Paik, S.J. Kim, H.J. Lee, J.Y. Jeon, J.Y. Lim, and T.R. Han. Movement of the hyoid bone and the epiglottis during swallowing in patients with dysphagia from different etiologies. Journal of Electromyography and Kinesiology, 18(2):329–335, 2008. 7→ page 3, 8, 11, 87, 121 [91] J.B. Palmer, E. Tanaka, and E. Ensrud. Motions of the posterior pharyngeal wall in human swallowing: a quantitative videofluorographic study. Archives of physical medicine and rehabilitation, 81(11):1520– 1526, 2000. 7→ page 121 [92] N. Papadopoulos, G. Lykaki-Anastopoulou, and E. Alvanidou. The shape and size of the human hyoid bone and a proposal for an alternative classification. Journal of anatomy, 163:249, 1989. 7→ page 5 [93] W.G. Pearson, S.E. Langmore, L.B. Yu, and A.C. Zumwalt. Structural analysis of muscles elevating the hyolaryngeal complex. Dysphagia, pages 1–7, 2012. 7→ page 118 [94] W.G. Pearson, S.E. Langmore, and A.C. Zumwalt. Evaluating the structural properties of suprahyoid muscles and their potential for moving the hyoid. Dysphagia, pages 1–7, 2011. 7→ page 12 [95] CC Peck, GEJ Langenbach, and AG Hannam. Dynamic simulation of muscle and articular properties during human wide jaw opening. Archives of Oral Biology, 45(11):963–982, 2000. 7→ page 118 101  Bibliography [96] CC Peck, AS Sooch, and AG Hannam. Forces resisting jaw displacement in relaxed humans: a predominantly viscous phenomenon. Journal of oral rehabilitation, 29(2):151–160, 2002. 7→ page 117 [97] J.P.V. Pelteret and BD Reddy. Computational model of soft tissues in the human upper airway. International Journal for Numerical Methods in Biomedical Engineering, 2012. 7→ page 16 [98] AL Perlman, PM Palmer, TM McCulloch, and DJ Vandaele. Electromyographic activity from human laryngeal, pharyngeal, and submental muscles during swallowing. Journal of Applied Physiology, 86(5):1663–1669, 1999. 7→ page 13 [99] A.L. Perlman, D.J. VanDaele, and M.S. Otterbacher. Quantitative assessment of hyoid bone displacement from video images during swallowing. Journal of speech and hearing research, 38(3):579, 1995. 7→ page 8 [100] G. Picinbono, H. Delingette, and N. Ayache. Nonlinear and anisotropic elastic soft tissue models for medical simulation. In Robotics and Automation, 2001. Proceedings 2001 ICRA. IEEE International Conference on, volume 2, pages 1370–1375. IEEE, 2001. 7→ page 40 [101] M.M. Porter, A.A. Vandervoort, and J. Lexell. Aging of human muscle: structure, function and adaptability. Scandinavian journal of medicine & science in sports, 5(3):129–142, 1995. 7→ page 5, 14, 15 [102] G. Pyka, E. Lindenberger, S. Charette, and R. Marcus. Muscle strength and fiber adaptations to a year-long resistance training program in elderly men and women. Journal of Gerontology, 49(1):M22– M27, 1994. 7→ page 87 [103] A.W. Rademaker, B.R. Pauloski, L.A. Colangelo, and J.A. Logemann. Age and volume effects on liquid swallowing function in normal women. Journal of Speech, Language, and Hearing Research, 41(2):275, 1998. 7→ page 13 [104] A. Ramos and JA Simoes. Tetrahedral versus hexahedral finite elements in numerical modelling of the proximal femur. Medical engineering & physics, 28(9):916–924, 2006. 7→ page 41 [105] J. Robbins, JW Hamilton, GL Lof, GB Kempster, et al. Oropharyngeal swallowing in normal adults of different ages. Gastroenterology, 103(3):823, 1992. 7→ page 4, 13, 14 102  Bibliography [106] B.C. Schiller, Y.G. Casas, B.L. Tracy, C.A. DeSouza, and D.R. Seals. Age-related declines in knee extensor strength and physical performance in healthy hispanic and caucasian women. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 55(12):B563–B569, 2000. 7→ page 16 [107] R. Schneiders. A grid-based algorithm for the generation of hexahedral element meshes. Engineering with Computers, 12(3):168–177, 1996. 7→ page 40 [108] W.S. Selbie, S.L. Gewalt, and C.L. Ludlow. Developing an anatomical model of the human laryngeal cartilages from magnetic resonance imaging. The Journal of the Acoustical Society of America, 112:1077, 2002. 7→ page 16, 116 [109] R. Shaker, M. Kern, E. Bardan, A. Taylor, ET Stewart, RG Hoffmann, RC Arndorfer, C. Hofmann, and J. Bonnevier. Augmentation of deglutitive upper esophageal sphincter opening in the elderly by exercise. American Journal of Physiology-Gastrointestinal and Liver Physiology, 272(6):G1518–G1522, 1997. 7→ page 87 [110] C.N. Shaw. Is ’hand preference’ coded in the hominin skeleton? an in-vivo study of bilateral morphological variation. Journal of human evolution, 2011. 7→ page 33 [111] DW Shaw, IJ Cook, M. Gabb, RH Holloway, ME Simula, V. Panagopoulos, and J. Dent. Influence of normal aging on oralpharyngeal and upper esophageal sphincter function during swallowing. American Journal of Physiology-Gastrointestinal and Liver Physiology, 268(3):G389–G396, 1995. 7→ page 14 [112] A. Sheffer, M. Etzion, A. Rappoport, and M. Bercovier. Hexahedral mesh generation using the embedded voronoi graph. In In Proceedings of the 7th International Meshing Roundtable. Citeseer, 1998. 7→ page 40 [113] H. Si. A quality tetrahedral mesh generator and three-dimensional delaunay triangulator. Weierstrass Institute for Applied Analysis and Stochastic, Berlin, Germany, 2006. 7→ page 39 [114] I. Sia, P. Carvajal, G.D. Carnaby-Mann, and M.A. Crary. Measurement of hyoid and laryngeal displacement in video fluoroscopic swal103  Bibliography lowing studies: Variability, reliability, and measurement error. Dysphagia, pages 1–6, 2011. 7→ page 10 [115] D.A. Skelton, C.A. Greig, J.M. Davies, and A. Young. Strength, power and related functional ability of healthy people aged 65–89 years. Age and ageing, 23(5):371–377, 1994. 7→ page 15 [116] D.A. Skelton, A. Young, C.A. Greig, K.E. Malbut, et al. Effects of resistance training on strength, power, and selected functional abilities of women aged 75 and older. Journal of the American Geriatrics Society, 43(10):1081, 1995. 7→ page 87 [117] B.C. Sonies, L.J. Parent, K. Morrish, and B.J. Baum. Durational aspects of the oral-pharyngeal phase of swallow in normal adults. Dysphagia, 3(1):1–10, 1988. 7→ page 13, 14 [118] B.C. Sonies, C. Wang, and D.J. Sapper. Evaluation of normal and abnormal hyoid bone movement during swallowing by use of ultrasound duplex-doppler imaging. Ultrasound in medicine & biology, 22(9):1169–1175, 1996. 7→ page 10, 121 [119] M. Sonomura, H. Mizunuma, T. Numamori, H. Michiwaki, and K. Nishinari. Numerical simulation of the swallowing of liquid bolus. Journal of Texture Studies, 2011. 7→ page 16 [120] J. Spiro, J.K. Rendell, and T. Gay. Activation and coordination patterns of the suprahyoid muscles during swallowing. The Laryngoscope, 104(11):1376–1382, 1994. 7→ page 13, 79 [121] I. Stavness, J. Lloyd, Y. Payan, S. Fels, et al. Towards speech articulation simulation with a dynamic coupled face-jaw-tongue model. International Seminar of Speech Production, 2011. 7→ page 17, 25 [122] I. Stavness, C. L. Ludlow, B. Chung, and S. Fels. Hyolaryngeal biomechanics modeling with intramuscular stimulation data. In 1st International Workshop on Dynamic Modeling of the Oral, Pharyngeal and Laryngeal Complex for Biomedical Applications (OPAL-2009), pages 73–78, Vancouver, BC, Canada, 2009. 7→ page iv, 4, 12, 66, 67, 79 [123] I.K. Stavness. Byte your tongue: a computational model of human mandibular-lingual biomechanics for biomedical applications. PhD thesis, University of British Columbia, 2010. 7→ page 67  104  Bibliography [124] Y. Su, KH Lee, and A. Senthil Kumar. Automatic hexahedral mesh generation for multi-domain composite models using a hybrid projective grid-based method. Computer-Aided Design, 36(3):203–215, 2004. 7→ page 39 [125] A. Tallgren and B. Solow. Hyoid bone position, facial morphology and head posture in adults. The European Journal of Orthodontics, 9(1):1–8, 1987. 7→ page 5, 87 [126] T.J. Tautges, T. Blacker, and S.A. Mitchell. The whisker weaving algorithm: A connectivity-based method for constructing all-hexahedral finite element meshes. International Journal for Numerical Methods in Engineering, 39(19):3327–3349, 1996. 7→ page 40 [127] N. Tayama, RW Chan, K. Kaga, and I.R. Titze. Geometric characterization of the laryngeal cartilage framework for the purpose of biomechanical modeling. The Annals of otology, rhinology, and laryngology, 110(12):1154, 2001. 7→ page 16 [128] G.J. Tortora and B. Derrickson. Principles of anatomy and physiology. Wiley, 2008. 7→ page 61, 114 [129] J.F. Tracy, J.A. Logemann, P.J. Kahrilas, P. Jacob, M. Kobara, and C. Krugler. Preliminary observations on the effects of age on oropharyngeal deglutition. Dysphagia, 4(2):90–94, 1989. 7→ page 13, 14 [130] L. Tsou and S. Fels. Biomechanical modeling of the external laryngeal structures. In International Symposium on Speech Sciences (ISSP’2011), pages 133–138, 2011. 7→ page iv [131] M. Uecker, S. Zhang, D. Voit, A. Karaus, K.D. Merboldt, and J. Frahm. Real-time mri at a resolution of 20 ms. NMR in Biomedicine, 23(8):986–994, 2010. 7→ page 9 [132] DJ Vandaele, AL Perlman, and MD Cassell. Intrinsic fibre architecture and attachments of the human epiglottis and their contributions to the mechanism of deglutition. Journal of anatomy, 186(Pt 1):1, 1995. 7→ page 10 [133] E. Vilkman and P. Karma. Vertical hyoid bone displacement and fundamental frequency of phonation. Acta oto-laryngologica, 108(12):142, 1989. 7→ page 13  105  [134] T.G. Wang, Y.C. Chang, W.S. Chen, P.H. Lin, and T.Y. Hsiao. Reduction in hyoid bone forward movement in irradiated nasopharyngeal carcinoma patients with dysphagia. Archives of physical medicine and rehabilitation, 91(6):926–931, 2010. 7→ page 8, 87, 121 [135] J.A. Weiss, B.N. Maker, and S. Govindjee. Finite element implementation of incompressible, transversely isotropic hyperelasticity. Computer methods in applied mechanics and engineering, 135(1):107–128, 1996. 7→ page 55 [136] P.L. Williams et al. Gray’s anatomy, volume 378. Churchill livingstone London, 1989. 7→ page 58, 59, 61, 110, 112, 113 [137] TN Willig, J. Paulus, S.G.J. Lacau, C. Béon, J. Navarro, et al. Swallowing problems in neuromuscular disorders. Archives of physical medicine and rehabilitation, 75(11):1175, 1994. 7→ page 5 [138] K. Yabunaka, H. Konishi, G. Nakagami, H. Sanada, S. Iizaka, S. Sanada, and M. Ohue. Ultrasonographic evaluation of geniohyoid muscle movement during swallowing: a study on healthy adults of various ages. Radiological physics and technology, pages 1–6, 2012. 7→ page 14, 121 [139] K. Yabunaka, H. Sanada, S. Sanada, H. Konishi, T. Hashimoto, H. Yatake, K. Yamamoto, T. Katsuda, and M. Ohue. Sonographic assessment of hyoid bone movement during swallowing: a study of normal adults with advancing age. Radiological physics and technology, 4(1):73–77, 2011. 7→ page 3, 10, 13, 14 [140] J. Yang, R. Chiou, A. Ruprecht, J. Vicario, LA MacPhail, and TE Rams. A new device for measuring density of jaw bones. Dentomaxillofacial Radiology, 31(5):313–316, 2002. 7→ page 116 [141] F.E. Zajac et al. Muscle and tendon: properties, models, scaling, and application to biomechanics and motor control. Critical reviews in biomedical engineering, 17(4):359, 1989. 7→ page 57, 117 [142] Y. Zu, Z. Yang, and A.L. Perlman. Hyoid displacement in posttreatment cancer patients: preliminary findings. Journal of Speech, Language, and Hearing Research, 54(3):813, 2011. 7→ page 8  106  Appendix A  Anatomy and Swallowing Physiology The appendix provides the background information on anatomy relevant to the topic of this thesis and a brief overview of the physiology of swallowing.  A.1  Anatomy  The process of swallowing involves a number of anatomical structures of the human body. Structures that have been discussed in this thesis, including components of the airway as well as many structures of the head and neck region, are described here.  A.1.1  Airway  The respiratory system allows the exchange of gasses for supplying the blood with oxygen. While the primary function is breathing, many of its structures are also involved in swallowing. The upper airway or the upper respiratory system comprises the nasal cavity, pharynx, larynx, and associated structures, whereas the lower respiratory system includes the trachea, bronchi, and lungs (see Figure A.1). Since the upper respiratory system is defined as comprising all structures above the glottic opening, the larynx can either be considered as a part of the upper or lower respiratory tract. Figure A.2 shows a more detailed illustration of the upper airway. Details on some of the structures are described in later sections.  A.1.2  Bones, Cartilages and Connective Tissues  Each of the rigid structures, i.e., bones and cartilages, included in the model are described in this section along with the associated connective tissues.  107  A.1. Anatomy  Figure A.1: The respiratory system. By Lord Akryl ( via Wikimedia Commons.  Figure A.2: Upper airway anatomy ( via Wikimedia Commons.  108  A.1. Anatomy Hyoid The hyoid bone is an unique bone that does not articulate with any other bone. As a part of the axial skeleton, it is suspended from the styloid process of the temporal bones by ligaments and muscles. It is located in the anterior neck above the larynx and forms the base of the tongue. The bone is closely coupled to the larynx by the thyrohyoid membrane (Figure A.3) and thyrohyoid muscle. Counting the genioglossus muscle, this horseshoe-shaped bone provides attachment sites for ten pairs of muscles in total. Six of them connect the hyoid bone with superior structures such as the skull and tongue: middle pharyngeal constrictor, hyoglossus, digastric, stylohyoid, geniohyoid, mylohyoid, and genioglossus muscle. Three of them connect the hyoid bone with inferior structures such as the sternum: thyrohyoid, omohyoid, and sternohyoid. This bone plays an important role in both speech production and swallowing. Laryngeal skeleton Nine cartilages form the laryngeal skeleton, which provides structural support to the larynx ( A.3). Three of them are unpaired: epiglottis, thyroid, and cricoid. The paired cartilages are the arytenoid, corniculate, and cuneiform cartilages. Among them, the thyroid and cricoid cartilages are the largest in volume and make up most of the laryngeal body. The two cartilages are coupled with the cricothyroid ligament and cricothyroid articulation. The epiglottis, unlike thyroid and cricoid, is a leaf-shaped elastic cartilage that folds down to protect the airway during the pharyngeal phase of swallowing. The three paired cartilages provide support to the internal soft structures of the larynx, such as the aryepiglottic folds and vocal folds. Skull The cranium is the part of the skull that protects the brain. The cranial bones include the frontal, parietal, temporal, occipital, sphenoid, and ethmoid bones (see Figure A.4). The mandible is the strongest facial bone and the only movable skull bone other than the ossicles (very small bones in the middle ear). It forms the lower jaw, whereas the maxilla forms the upper jaw.  109  A.1. Anatomy  (a) Anterior view  (b) Posterior view  Figure A.3: Cartilages of the larynx. Note that the paired cuneiform cartilages are not drawn here. Adapted from (a) CC by Olek Remesz via Wikimedia Commons and (b) Williams et al, 1989 [136].  Figure A.4: Side view of the skull. Adapted from Williams et al, 1989 [136].  110  A.1. Anatomy Sternum and bones of the shoulder The sternum is a long flat bone of the chest and consists of three parts (Figure A.5. From top to bottom, they are the manubrium, gladiolus, and xiphoid process. The human shoulder is made of the clavicle, scapula, and humerus. The cavicle, also known as the collar bone, is a long curved bone that runs horizontally just above the scapula. The scapula is a triangleshaped bone, also known as the shoulder blade, that connects the clavicle and the humerus, which is the bone of the upper arm. Trachea The trachea is a tube located anterior to the esophagus. It extends from the larynx by the cricotracheal ligament and provides a passageway for air. It consists of the 16–20 tracheal cartilages (or tracheal rings) and annular ligaments. The tracheal cartilages are incomplete rings that are shaped like the letter ”C,” whereas the annular ligaments are circular bands of fibrous tissue that connect the cartilages with each other.  Figure A.5: Bones of the shoulder. Adapted from illustration by National Cancer Institute via Wikimedia Commons.  111  A.1. Anatomy  A.1.3  Muscles  Muscles included in the model are introduced here. Extrinsic Laryngeal Muscles There are two groups of muscles associated with the larynx: intrinsic and extrinsic. The former connect the larynx or the closely coupled hyoid bone to the surrounding structures. They can be further divided into the suprahyoid and infrahyoid muscle groups. The muscles are illustrated in Figure A.6 and summarized in Table A.1. Suprahyoid Muscles The suprahyoid muscles are extrinsic laryngeal muscles that connect to structures superior to the larynx. They are labeled with red text in Figure A.6. Their individual functions are listed in Table A.1. In general, they elevate the hyoid bone. Infrahyoid Muscles The infrahyoid muscles are extrinsic laryngeal muscles that connect to structures inferior to the larynx. They are labeled with blue text in Figure A.6. Their individual functions are listed in Table A.1. In general, they depress the hyoid bone.  Figure A.6: Extrinsic laryngeal muscles. Adapted from Williams et al, 1989 [136].  112  A.1. Anatomy Pharyngeal Constrictors The pharyngeal constrictors are muscles of the pharynx that form the back wall of the throat. They are sheet-like muscles that originate from the larynx, hyoid, and lateral-anterior parts of the skull. The muscles from both sides join at the posterior mid-line and form the pharyngeal raphe (see Figure A.7). They constrict the pharynx and convey the food bolus downward into the esophagus when one swallows.  Figure A.7: Pharyngeal constrictor muscles. Adapted from Williams et al, 1989 [136].  113  INSERTION  ACTION  Hyoid bone via an intermediate tendon Hyoid bone via an intermediate tendon Hyoid bone  Mylohyoid  Hyoid bone Hyoid bone  Elevates hyoid bone and depresses mandible, as in opening the mouth Elevates hyoid bone and depresses mandible, as in opening the mouth Elevates hyoid bone, draws hyoid bone and tongue anteriorly, and depresses mandible. Elevates hyoid bone and floor of mouth and depresses mandible. Elevates hyoid bone and draws it posteriorly.  Intermediate tendon  Depresses hyoid bone.  Intermediate tendon  Depresses hyoid bone.  Hyoid bone Thyroid cartilage Hyoid bone  Depresses hyoid bone. Depresses thyroid cartilage. Elevates thyroid cartilage and depresses hyoid bone.  Mandible  Stylohyoid Temporal bone Infrahyoid Muscles Omohyoid, Scapula inferior belly Omohyoid, Hyoid bone superior belly Sternohyoid Clavicle and sternum Sternothyroid Sternum Thyrohyoid Thyroid cartilage  Table A.1: Extrinsic laryngeal muscles. Based on [128].  A.1. Anatomy  MUSCLE ORIGIN Suprahyoid Muscles Digastric, Mandible anterior belly Digastric, Styloid process of posterior belly the temporal bone Geniohyoid Mandible  114  A.2. Swallowing Physiology  A.2  Swallowing Physiology  The normal swallowing mechanism can be broken down into three phases, and each is described below. See Figure 2.1 for mid-sagittal views of the upper airway at each stage of the swallowing cycle.  A.2.1  Oral Phase  The oral phase first involves preparing the food for swallowing—the food is moistened with saliva and masticated to form a bolus. The bolus is then pushed from the front of the mouth to the back of the throat.  A.2.2  Pharyngeal Phase  Once the bolus reaches the pharynx, it triggers the pharyngeal swallowing reflex, which initiates a series of involuntary motions that further move the bolus into the esophagus. During this phase, breathing is stopped to prevent the bolus from entering the airway. This is enforced by the closure of epiglottis, vestibular folds, and vocal folds. The hyoid and larynx are elevated and the UES is opened allowing the bolus to enter the esophagus.  A.2.3  Esophageal Phase  The food bolus is propelled down the esophagus by sequential contraction and relaxation of the muscle wall in the last phase of swallowing. Once it reaches the end of the esophagus, the lower esophageal sphincter relaxes and the bolus is pushed into the stomach.  115  Appendix B  Mechanical Properties B.1  Rigid Bodies  A rigid body in the ArtiSynth modeling environment is defined simply by its shape and mass. Table B.1 provides a list of rigid body masses utilized in our model. Name Cricoid Hyoid Thyroid Tracheal Ring 1 2 3  Mass[g] 5.261 2.912 8.521 3.783  From [108]. Based on bone density 1.85 g/cm3 [140]. Based on density 1.14 g/cm3 [43]. Table B.1: Rigid body property  B.2  Finite Element Models  FEM was implemented either using a hyperelastic Mooney–Rivlin material or a linear material. Rayleigh damping coefficients of α = 40s−1 and β = 0.5msec were applied to all the FEM models to provide appropriate damping to the system.  116  B.3. Muscle Model  B.2.1  Mooney Rivlin Material  A fifth-order Mooney–Rivlin model was utilized for muscles that were modeled by FEM. The strain-energy function W is defined by W =c10 (I1 − 3) + c01 (I2 − 3) + c20 (I1 − 3)2 + c11 (I1 − 3)(I2 − 3) + c02 (I2 − 3)2 + 1 Kln2 J 2  (B.1)  where I1 is the first invariant of the right Cauchy–Green strain tensor, I2 is the second invariant of the right Cauchy–Green strain tensor, J is the determinant of the elastic deformation gradient, and K is the bulk modulus. Note that the last term of this equation, 12 Kln2 J, can be considered as a penalty function for generating restoring forces for maintaining constant volume. If the volume equals the rest volume, the determinant of the deformation gradient J = 1 and the penalty function would be zero. The Mooney–Rivlin constants [Pa] utilized by our model were based on [11], where c10 =1037, c20 =486, and c20 = c11 = c02 = 0. A bulk modulus of 10370 Pa was assigned to each muscle FEM model.  B.2.2  Linear Material  Connective tissues such as tendons and ligaments were modeled as linear materials characterized by their Young’s modulus and Poisson’s ratio. Table B.2 provides a list of the values used in the model. Due to the lack of data in the literature, these values were set such that appropriate behaviors were observed when the model was in motion.  B.3  Muscle Model  The active springs were modeled as Hill-type actuators with a maximum length of 150% of the rest length [141]. The same formulation was used in [64, 96] for modeling jaw muscles. The maximum force of each muscle is listed in Table B.3 along with the mass of the FEM mesh superimposed by the springs. Note that the sylohyoid was modeled as a single pair of springs without an FEM mesh; therefore, no mass was associated with it.  117  B.3. Muscle Model  Name  Mass[g]  Cricothyroid ligament Digastric intermediate tendon Omohyoid intermediate tendon Thyrohyoid membrane  0.3121 0.0675 0.0106 1.3751  1  Young’s Modulus [Pa] 1.2E8 1.2E10 1.2E10 5E5  Poisson’s ratio 0.48 0.48 0.48 0.48  Approximated from density 1.06 g/cm3 [43]. Table B.2: Connective tissue properties, linear FEM material  Muscle Digastric anterior Digastric posterior Geniohyoid Mylohyoid Omohyoid inferior Omohyoid superior Sternohyoid Sternothyroid Stylohyoid Thyrohyoid 1 2 3  Mass[g] 2.371 2.531 2.211 5.201 1.92 1.05 6.15 4.54 1.561  PCSA[cm2 ] 0.551 0.641 0.461 1.251  0.271 0.511  Fmax [N]3 22 25.6 18.4 50 20 20 502 502 10.8 20.4  From [93]. From [42]. Fmax = P CSA × 40N/cm2 [95] Table B.3: Muscle properties  118  B.4. Spring Network  B.4  Spring Network  Passive springs were modeled as simple massless spring–damper systems defined by their stiffness and damping coefficient. See Table B.4 for the values used in the model. Again, these values were approximated such that the system behaved appropriately during simulation. Name Annular ligaments of trachea Cricotracheal ligament Digastric intermediate tendon sheath Omohyoid intermediate tendon sheath Pharyngeal constrictor, inferior Pharyngeal constrictor, middle  Stiffness [N/m] 50 50 200 200 50 50  Damping 0.5 0.5 0.5 0.5 0.5 0.5  Table B.4: Spring network properties  119  Appendix C  Additional Material  120  Source  Module  Dodds(1988)[25]  Bolus [ml]  Subject  Age  Anterior Disp. [cm]  Elevation Disp. [cm]  Thick liquid  dry, 2, 5, 10, 15, 20  15 healthy  20-82  1.15-1.69  1.11-1.51  Liquid, paste  2, 5, 10, 20  10 healthy  18-36  1.07-1.44  1.01-1.45  -  Ishida(2002)[53]  Videofluorography x,y-axes with respect to the cervical spine Videofluorography x,y-axes with respect to the cervical spine Videofluorography x-axis = occlusal plane  Overall Disp. [cm] -  Liquid, chewed solids  8  12 healthy  20-28  1.18-1.29  0.65-1.16  -  Kendall(2001)[58]  Videofluorography -  Liquid  1, 20  18-62 67-83 >65  -  -  1.19-2.27 1.63-2.47 0.98-2.19  Kim(2008)[60]  Videofluorography  Liquid  5, 10  Leonard(2000)[66]  Videofluorography  Liquid  1, 3, 20  Logemann(2000)[70]  Videofluorography  Liquid  1, 10  Logemann(2002)[71]  Videofluorography  Liquid  1, 10  Mays(2009)[79]  Videofluorography  Liquid  12, 24  Paik(2008) [90]  y-axis = anterior Videofluorography corners of C2 and C4  Liquid  5  60 healthy 23 healthy 46 for 1ml and 16 for 20ml, with dysphagia 20 healthy 20 healthy 60 healthy 8 healthy men 8 healthy men 8 healthy women 8 healthy women 12 healthy 9 healthy 7 stroke with dysphagia 3 myopathy with dysphagia 6 healthy 3 healthy  21-51 70-87 18-73 21-29 80-94 21-29 80-93 20-29 60.2±4.1 63±4.4  1.62-1.80 0.99-1.17 0.728 0.514 1.011 1.032 1.229-5.717 1.5 1.1  1.58-1.64 1.45-1.53 1.933 0.728 1.294 1.54 0.532-3.401 1.3 1.2  63±6.2  0.4  0.8  33.3±12.8 -  1.28 0.81-0.96  1.29 0.54-0.73  1.13-1.29  10 healthy 33 irradiated with dysphagia 10 healthy 10 healthy 10 healthy  53.9±4.8 55.5±8.8  1.65 0.85  1.38 1.25  2.23 1.58  20-39 40-59 60-79  -  1.66 1.47 1.34  -  y-axis = anterior corners of C2 and C4 y-axis = anterior surface of C2 and C4 y-axis = anterior corners of C2 and C4 x-axis = occlusal plane  Palmer(2000)[91] Sonies(1996)[118]  Videofluorography Ultrasound Triangular duplex-doppler fitting imaging  Liquid 15 consecutive liquid swallow  dry, 10, 20 10  Wang(2010)[134]  Videofluorography -  Thick liquid  5  Yabunaka(2012)[138]  Ultrasonography  Liquid  5  -  trajectory  NPC  1.69-2.16 -  Table C.1: Hyoid displacement reported by past studies. Note that the displacements included above are ranges of mean values across different bolus volumes or subject groups.  Appendix C. Additional Material  Bolus Type  Dantas(1990)[23]  Rotation Correction  121  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items