In silico estimation of labral deformation in hips with cam femoroacetabular impingement syndrome during gait and pivoting activities

Cam femoroacetabular impingement syndrome (FAIS) is a hip condition characterized by motion or position-related pain arising from morphological abnormalities of the proximal femur. Cam morphology is theorized to alter the mechanical environment of the joint, causing conflict between the acetabular labrum and proximal femur. Soft tissue damage in the anterosuperior labrum is often observed clinically, and research suggests that cam FAIS is major underlying cause for many cases of hip osteoarthritis (OA), formally deemed idiopathic. Improving our understanding of impingement and its relationship to joint morphology and motion may improve patient care for those with cam FAIS. Computational models are frequently used to study the pathomechanisms of impingement, otherwise impossible to study in vivo, but many models apply assumptions or simplifications that are not appropriate for describing the complex interaction of morphology and motion in cam FAIS. Models with patient-specific inputs address this limitation but require time-intensive setup and optimizations, rendering them less feasible for clinical applications. To balance patient-specificity and computational simplicity, we adapted a modeling technique, used by several research groups to study knee and ankle mechanics, which we refer to as "soft tissue overlap" (STO) modeling. In the first portion of this thesis, we used STO modeling to investigate how labral contact mechanics may be altered in cam FAIS compared to healthy controls in the acetabular labrum during four activities of daily living (internal/external pivots, level/inclined walking). No statistically significant differences were found during pivoting activities, but trends suggesting greater contact area and strain in the acetabular labrum of the cam FAIS cohort were found during the walking activities. In the second portion of this thesis, we investigated the sensitivity of STO model predictions by introducing perturbations in mesh geometry and kinematics and comparing an STO model to a finite element model (herein considered a reference standard for modeling of hip contact mechanics). With both portions of this thesis, we hope to offer new insight into the mechanical iv mysteries surrounding cam FAIS and provide a framework for a new computational methodology for future studies to improve upon.

IN SILICO ESTIMATION OF LABRAL DEFORMATION IN HIPS WITH CAM FEMOROACETABULAR IMPINGEMENT SYNDROME DURING GAIT AND PIVOTING ACTIVITIES by Lindsay Lea Schuring A thesis submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Master of Science Department of Bioengineering The University of Utah December 2019 Copyright © Lindsay Lea Schuring 2019 All Rights Reserved The University of Utah Graduate School STATEMENT OF THESIS APPROVAL The thesis of Lindsay Lea Schuring has been approved by the following supervisory committee members: Andrew E. Anderson , Chair August 7th, 2019 Date Approved Robert S. MacLeod , Member August 7th, 2019 Date Approved Lucas H. Timmins , Member August 7th, 2019 Date Approved and by David W. Grainger the Department/College/School of and by David B. Kieda, Dean of The Graduate School. , Chair/Dean of Bioengineering ABSTRACT Cam femoroacetabular impingement syndrome (FAIS) is a hip condition characterized by motion or position-related pain arising from morphological abnormalities of the proximal femur. Cam morphology is theorized to alter the mechanical environment of the joint, causing conflict between the acetabular labrum and proximal femur. Soft tissue damage in the anterosuperior labrum is often observed clinically, and research suggests that cam FAIS is major underlying cause for many cases of hip osteoarthritis (OA), formally deemed idiopathic. Improving our understanding of impingement and its relationship to joint morphology and motion may improve patient care for those with cam FAIS. Computational models are frequently used to study the pathomechanisms of impingement, otherwise impossible to study in vivo, but many models apply assumptions or simplifications that are not appropriate for describing the complex interaction of morphology and motion in cam FAIS. Models with patient-specific inputs address this limitation but require time-intensive setup and optimizations, rendering them less feasible for clinical applications. To balance patient-specificity and computational simplicity, we adapted a modeling technique, used by several research groups to study knee and ankle mechanics, which we refer to as “soft tissue overlap” (STO) modeling. In the first portion of this thesis, we used STO modeling to investigate how labral contact mechanics may be altered in cam FAIS compared to healthy controls in the acetabular labrum during four activities of daily living (internal/external pivots, level/inclined walking). No statistically significant differences were found during pivoting activities, but trends suggesting greater contact area and strain in the acetabular labrum of the cam FAIS cohort were found during the walking activities. In the second portion of this thesis, we investigated the sensitivity of STO model predictions by introducing perturbations in mesh geometry and kinematics and comparing an STO model to a finite element model (herein considered a reference standard for modeling of hip contact mechanics). With both portions of this thesis, we hope to offer new insight into the mechanical mysteries surrounding cam FAIS and provide a framework for a new computational methodology for future studies to improve upon. iv TABLE OF CONTENTS ABSTRACT ...................................................................................................................................... iii LIST OF TABLES ............................................................................................................................ vi LIST OF FIGURES ......................................................................................................................... vii ACKNOWLEDGEMENTS .............................................................................................................. viii Chapters 1. INVESTIGATION OF LABRAL CONTACT MECHANICS IN CAM FEMOROACETABULAR IMPINGEMENT SYNDROME……………….……………………………………………………………1 Introduction and Background ............................................................................................... 1 Motivation ........................................................................................................................ 1 Healthy Hip Anatomy and Physiology ............................................................................. 2 Osteoarthritis and Cam FAIS........................................................................................... 3 Computational Modeling of Hip Contact Mechanics........................................................ 6 Objective .......................................................................................................................... 8 Methods ............................................................................................................................... 9 Participant Recruitment ................................................................................................... 9 Participant Imaging .......................................................................................................... 9 Model Construction and Kinematics .............................................................................. 10 Contact Analysis ............................................................................................................ 12 Results ............................................................................................................................... 14 Internal and External Pivots ........................................................................................... 14 Level and Inclined Walking ............................................................................................ 14 Discussion ......................................................................................................................... 15 2. SENSITIVITY STUDIES OF SOFT TISSUE OVERLAP MODELING AND COMPARISON TO FINITE ELEMENT MODELING..………………………………………………………...............……..30 Introduction, Motivation, and Objectives ........................................................................... 30 Methods ............................................................................................................................. 32 Chondrolabral Boundary Sensitivity .............................................................................. 32 Kinematic Sensitivity ...................................................................................................... 33 Finite Element Model Comparison ................................................................................ 33 Results ............................................................................................................................... 34 Chondrolabral Boundary Sensitivity .............................................................................. 34 Kinematic Sensitivity ...................................................................................................... 35 Finite Element Model Comparison ................................................................................ 36 Discussion ......................................................................................................................... 36 3. SUMMARY, CONCLUSIONS, AND IMPACT…………...………………………………………….48 REFERENCES .............................................................................................................................. 50 LIST OF TABLES Tables 1.1. Participant demographics ....................................................................................................... 29 1.2. ISB standard hip coordinate system. ...................................................................................... 29 2.1. Material definitions and parameters of FE model ................................................................... 47 2.2. Contact definitions of FE model ............................................................................................. 47 2.3. Kinematic sensitivy results. .................................................................................................... 47 LIST OF FIGURES Figures 1.1. Schematics of the human hip joint. ........................................................................................ 19 1.2. Schematics of the hip from a superior view.. .......................................................................... 20 1.3. Depiction of impingement ....................................................................................................... 21 1.4. Methods flowchart. ................................................................................................................. 22 1.5. STO model methodological schematic. .................................................................................. 23 1.6. Internal/External pivots at end stance .................................................................................... 24 1.7. Mean and max walking trial results ........................................................................................ 25 1.8. Dynamic level walking results................................................................................................. 26 1.9. Dynamic inclined walking results ............................................................................................ 27 1.10. Selected comparisons of t-test and SPM results .................................................................. 28 2.1. Part I of chondrolabral boundary computational methods ..................................................... 40 2.2. Part II of chondrolabral boundary computational methods .................................................... 40 2.3. Chondrolabral boundary sensitivity results (±10 mm) ............................................................ 41 2.4. Chondrolabral boundary sensitivity results (±4 mm) .............................................................. 41 2.5. Translational error results in kinematic sensitivity analysis .................................................... 42 2.6. Rotational error results in kinematic sensitivity analysis ........................................................ 43 2.7. Qualitative comparison of contact area, STO vs. FE modeling .............................................. 44 2.8. Qualitative comparison of contact strains, STO vs. FE modeling ........................................... 45 2.9. Quantitative comparison of contact area and strains, STO vs. FE modeling .......................... 46 ACKNOWLEDGEMENTS I would like to express my sincere gratitude and appreciation for my advisor Dr. Andy Anderson for the guidance he has given and the trust he has placed in me, and to my other committee members, Dr. Rob MacLeod and Dr. Luke Timmins, for their mentorship throughout my time at the University of Utah. I would also like to thank my fellow lab members and associated colleagues for their guidance and support: Dr. Amy Lenz, Dr. Penny Atkins, Dr. Koren Roach, Dr. Keisuke Uemura, Dr. Joe Mozingo, Dr. Nic Fiorentino, Dr. Steve Maas, Dr. Bo Foreman, Dr. Jeff Weiss, Dr. Cara Lewis, Jocelyn Todd, Klevis Aliaj, Hema Sulkar, Chris Kolz, Carolyn Taylor, Anndee Neuman, Dylan Blair, Trevor Hafer, Sam Colby, Joe Hartle, YoungJae Shin, and Rich Lisonbee. To my friends and family, thank you for helping me maintain sanity throughout my endeavors: Devon Wooley, David Jiang, Annie Rowley, Paisley Tarboton, Tivon Semnani, Melanie Stokes, Katie Stokes, Nanea Phenicie, Karina Angulo, Orane Spence, Grace Heaps Jensen, Tania Helsten Arguello, Kody Kendall, Tanner Gilliland, Aubrey, Nick, Ian, Nate, Rocky, Joey, Mom, Dad, Grandma Spendlove, and Grandma and Grandpa Schuring. Finally, I would like to acknowledge all the sources of funding that have supported me financially throughout my career in the Harold K. Dunn, MD, Orthopaedic Research Laboratory. CHAPTER 1 INVESTIGATION OF LABRAL CONTACT MECHANICS IN CAM FEMOROACETABULAR IMPINGEMENT SYNDROME Introduction and Background Motivation A growing body of evidence suggests that many cases of osteoarthritis (OA) originally presumed idiopathic (without decipherable cause such as injury) can be attributed to an underlying pathology called cam femoroacetabular impingement syndrome (FAIS) (Abraham et al., 2013; Barros et al., 2010; Ganz et al., 2008; Groh and Herrera, 2009; Harris, 1986; Henak et al., 2011; Knight et al., 2017). Cam FAIS is a hip condition characterized by motion or position-related pain arising from abnormal bone morphology of the proximal femur, with reduced femoral head-neck offset and femoral head sphericity (Griffin et al., 2016). Cam FAIS is thought to alter the mechanical environment of the joint, as patients often present with acetabular cartilage delamination and/or tears to the acetabular labrum along with symptoms of clicking, popping or locking of the hip (Griffin et al., 2016). Clinical findings dating back to the mid-1980s support the theory that the morphological deformities associated with cam FAIS are a risk factor for OA, yet the pathogenesis of OA in patients with FAIS has not been firmly established (Ganz et al., 2008; Harris, 1986; Zhang et al., 2015). The clinical definition and diagnostic requirements for cam FAIS have only recently been described, gaining recognition in the early 2000s and formally standardized in 2016 (Ganz et al., 2003; Griffin et al., 2016; Ito et al., 2001). Because ambiguity still surrounds cam FAIS and its mechanisms of joint degradation/damage, treatment options continue to evolve. Many patients elect to undergo surgery for management of their symptoms; short- to midterm outcomes are largely a success in terms of pain relief and return to activity. However, 5-10% of patients require 2 revision surgery (Levy et al., 2016) and 10-15% progress to joint replacement within 4 years (Larson et al., 2014, 2012, 2008). The major motivation of this thesis, thus, was to investigate if and how labral contact mechanics are altered by cam FAIS, which may guide future research and treatment of the condition and aid in the prevention OA. Healthy Hip Anatomy and Physiology The hip is a synovial ball and socket joint, the ball component being the femoral head and the socket being the cup-like acetabulum of the pelvis (Figure 1.1). Both surfaces are lined with hyaline cartilage between 1-4 mm in thickness. By weight, articular cartilage is composed of 7085% interstitial fluid, with the remaining solid matrix primarily composed of type II collagen, giving the cartilage strength under tension and shearing deformations (Griffin et al., 2016; Groh and Herrera, 2009; Klennert et al., 2017). The mechanical behavior of cartilage is nonlinear and ratedependent as a result of fluid-solid interactions and its viscoelastic solid phase behavior. Under high-frequency loading (such as in walking), fluid does not have time to exude from the cartilage solid matrix; in this case, the mechanical behavior of cartilage can be approximated as an incompressible hyperelastic material (Abraham et al., 2013; Anderson et al., 2008; Knight et al., 2017; Todd et al., 2018). The micro- and macrostructures of cartilage allow load to be dispersed across the joint and provide surfaces with low coefficients of friction for articulation (Allen et al., 2010; Bharam, 2006). Load dispersion and smooth articular surfaces are crucial for healthy hip function, as the hip affords large ranges of motion while under considerable loading, e.g., 1.5 - 5 times bodyweight during activities such as walking or running (Bergmann et al., 2001). The acetabulum is also lined with a fibrocartilaginous tissue, the acetabular labrum, thought to aid in additional load dispersion and joint stability (Figure 1.1). The labrum is approximately 1-3 mm thick radially (Bharam, 2006) and follows the ridge of the acetabulum in a circular arc to form a lip surrounding the femoral head (Bharam, 2006; Groh and Herrera, 2009). In its cross section, the labrum is approximately triangular in shape with a height of 4-9 mm from the apex of the acetabular ridge to the labral apex (Karns et al., 2016) (Figure 1.1). Acetabular cartilage connects to the labrum through a nearly continuous transition zone called the chondrolabral junction or 3 boundary (Lewis and Sahrmann, 2006). The labrum can be separated into three distinct layers: (1) the articular surface, with superficial nerve endings and vascularization covered by a meshwork of thin fibrils; (2) the internal layer, with lamella-like fibrils; and (3) the final layer composed primarily of circumferential type I collagen fibers (Alzaharani et al., 2014; Petersen et al., 2003). The unique structure of labrum is thought to serve many functions, including the prevention of lateral femoral head translation, load dispersion from peripheral cartilage, nociception, and proprioception (Alzaharani et al., 2014; Banerjee and Mclean, 2011; Ferguson et al., 2000; Groh and Herrera, 2009; Harris et al., 2012; Todd et al., 2018). Altogether, the functions likely protect the hip joint from chondral damage (Groh and Herrera, 2009). Pathologies characterized by anatomical abnormalities such as cam FAIS are thought to disrupt normal joint articulation and increase loads experienced at the labrum, leaving acetabular cartilage susceptible to injury, degradation, and OA (Henak et al., 2011; Ito et al., 2001; Leunig et al., 2003; Safran et al., 2011). Osteoarthritis and Cam FAIS Osteoarthritis (OA) is among the leading causes of disability across the globe (Cross et al., 2014) and is the most common musculoskeletal disease in the US, affecting an estimated 30 million Americans presently (Cross et al., 2014; Dahlberg et al., 2016). OA reportedly cost the US $16.5 billion in 2013, making it the second most costly health condition treated in US hospitals that year (Martin et al., 2016). This irreversible degenerative joint disease is marked by the breakdown of cartilaginous tissues and their typical metabolic processes, causing inflammation, swelling, bone degradation, and pain. One of the most commonly affected joints is the hip, with one in four people at risk to develop hip OA over the span of a lifetime (Murphy et al., 2010). Once end-stage hip OA is reached, the only treatment option left for patients to regain hip mobility is total hip arthroplasty (THA), in which the native hip is replaced with a prosthesis. The number of THA procedures has been projected to increase 70% between the years 2014-2030 (Sloan et al., 2018), with more procedures likely to be performed on people under the age of 65 than ever before (National Center for Health Statistics, 2017). Young age coupled with the ever-increasing average life expectancy means that more patients can expect to outlive their THA prostheses and may require expensive 4 revision surgeries, which cost on average 76% more than primary THA procedures (National Center for Health, 2017; Sloan et al., 2018; Weber et al., 2018). Despite THA being deemed one of the most successful interventions in modern medicine, our understanding of etiological factors involved in the development of OA has not come as far (Knight et al., 2011). By addressing factors that may leave some more at risk for the development of OA, the native hip joint may be preserved longer, and costs associated with THA may be delayed or prevented entirely. A growing body of evidence suggests that many cases of osteoarthritis (OA) originally presumed idiopathic (without decipherable cause such as injury), especially in young active adults, can be attributed to underlying structural abnormalities of the pelvis and/or femur (Abraham et al., 2013; Barros et al., 2010; Ganz et al., 2008; Groh and Herrera, 2009; Harris, 1986; Henak et al., 2011; Knight et al., 2017). The first study to provide clinical data in support of this conjecture was Harris et al. in the mid-1980s, reporting that abnormal bone morphologies were present in 90% of patients with idiopathic hip OA; one of the most common abnormalities seen in their patients was excessive bone on the proximal femur, termed a “pistol grip deformity.” It was not until the early 2000s that Ganz et al. would formally coin the term “cam FAI” to describe the mechanical effects of such abnormal femoral morphology seen by Harris’s group (Ganz et al., 2003, 2001). By 2010, with the pathology of cam FAIS described, Barros et al. performed a study similar to that of Harris’s group, finding that 80% of hips with end-stage “idiopathic” OA showed signs of the femoral deformities seen in cam FAIS, compared to 30% in healthy controls of similar demographics (Barros et al., 2010; Harris, 1986). FAIS is a painful motion-related condition of the hip, in which excessive bone growth introduces mechanical conflict between tissues of the acetabulum and femoral head/head-neck junction (Figure 1.2) (Griffin et al., 2016). Cam FAIS is one of two types of FAIS (Figure 1.2). Excess bone growth can be located on the acetabular rim (pincer FAIS), or femoral head and head-neck junction (cam FAIS), but it is common to see both types together (combined FAIS). In order to be diagnosed with either type of FAIS, both anatomical deformities and symptoms of pain related to joint motion or positioning must be present (Griffin et al., 2016). “FAI morphology” describes anatomical deformities that are seen in FAIS, but it is important to note that this is distinct from 5 FAIS, as it is possible to have structural hip deformities without clinical signs of impingement or symptoms of pain (Griffin et al., 2016). The association of cam FAIS and hip OA is much stronger than pincer FAIS, and isolated cam morphology is more common than isolated pincer morphology; therefore, cam FAIS serves as the focus of this thesis. The etiology of cam morphology has been debated, but the higher prevalence of cam morphology in athletes and active adults suggests that the cam lesion likely develops during adolescence as a response to repeated, intensive loading of the joint, prior to closure of the femoral head physis (growth plate) (Anderson et al., 2016; Economopoulos et al., 2014; Kapron et al., 2015, 2011). Although the relationship of FAI morphology, FAIS, and the development of OA from FAIS is complex, we know that the unique morphological abnormalities of FAIS do relate to specific damage characteristics in the hip (Beck et al., 2005; Dahlberg et al., 2016; Ito et al., 2001; Leunig et al., 2003). Compared to pincer morphology (and other pathomorphologies), the cam lesion (Figure 1.3) is associated with more rapid and severe joint damage and degeneration (Groh and Herrera, 2009; Ito et al., 2001; Leunig et al., 2003). Cam-dominant forms of FAIS often display focal damage primarily located in the anterosuperior region of the joint, where the acetabular cartilage and labrum are typically thinner than in other regions (Banerjee and Mclean, 2011; Clohisy et al., 2009; Groh and Herrera, 2009). Separation between the labrum and acetabular cartilage is another defining characteristic of cam-dominant FAIS joint damage and is thought to be the primary reason for accelerated joint degradation in cam FAIS (Banerjee and Mclean, 2011) (Figure 1.3). McCarthy et al. (2001) found that 73% of patients with tears, fraying, or defibrillation to the acetabular labrum also had acetabular chondral damage. Additionally, isolated labral tears were more often found in younger patients, whereas labral tears in combination with chondral damage were more often found in older patients, meaning labral tears likely precede chondral damage (McCarthy et al., 2001). Compromised labral tissues may disrupt load dispersion across the joint, with stresses focused on the peripheral acetabular cartilage. Finite element (FE) studies have indicated that the removal of the labrum results in greater observed stresses and strain at the periphery of the acetabular cartilage (Ferguson et al., 2000; Todd et al., 2018). Altogether, both clinical studies and computational modeling predictions indicate that the labral tears characteristic 6 of cam FAIS leave the acetabular cartilage more susceptible to degeneration or delamination (Ferguson et al., 2000; Lewis and Sahrmann, 2006; McCarthy et al., 2001; Todd et al., 2018). Three common treatment pathways can be taken to prevent or repair labral damage due to cam FAIS. The most conservative pathway involves physical therapy, targeting muscular stability and promoting alterations to habitual motions in critical activities (Griffin et al., 2016). If physical therapy is unsuccessful, the second course of action includes the addition of antiinflammatory medications, or corticosteroid injections. If still ineffective, surgical intervention, involving repair to soft tissue damage, and/or resection of the cam lesion, may be necessary. Labral and chondral repair outcomes could be improved; some studies suggest that upwards of 20% of patients are worse off with treatment than no treatment (Groh and Herrera, 2009). Resection surgeries can also be insufficient because treatment outcomes are understudied, leading clinicians either remove too much or too little bone, leaving patients susceptible to cortical bone failure (rare), increased joint instability (common), or continued impingement that requires revision surgeries (common). Successful treatment of FAIS relies on the accurate characterization of its pathoanatomic nature, including the location and severity of impingement from the condition. By providing insightful biomechanical models of FAIS, we can begin to address these needs and improve our understanding of the relation between the acetabular labrum and mechanisms of damage (Banerjee and Mclean, 2011). Computational Modeling of Hip Contact Mechanics A standard modeling method for predicting the behavior of mechanical systems, including the hip, is finite element (FE) modeling. FE models simulate mechanical systems (defined by forces, displacements, velocities, constitutive equations, etc.) using iterative numerical methods to arrive at a solution that satisfies mechanical laws and bounds defined within the system (FEBio User Manual, Software Suite v2.2.0, University of Utah, Salt Lake City). In the context of biomechanics of the hip, FE models have provided a wealth of information that would have otherwise been impossible or extremely difficult to obtain in vivo or in cadaver studies. However, FE models require an extensive number of computational optimizations that can be incredibly time- 7 intensive, making them difficult to run multiple simulations for broader studies with more participants. To overcome this, other modeling approaches have been used to either simplify model setup or model computations and optimizations. Instead of using volumetric meshes as in FE models, discrete element analysis (DEA) models utilize springs to represent material deformations. Abraham et al. (2013) reported that solution times for their DEA models took ~7 seconds, much faster than a related FE model (~65 minutes, after many hours of model debugging and tuning). However, DEA still requires significant setup time to create springs from meshes, and the number of springs chosen for the model typically results in reduced resolution (spring density). Results are also difficult to visualize because of this resolution reduction, an important aspect of modeling critical for clinical interpretation. Bone collision models are another approach that have been applied to estimate range of motion (ROM) and impingement locations in hip joint modeling (Audenaert et al., 2011; Bedi et al., 2011; Kubota et al., 2017; Sochacki et al., 2018). Here, generalized kinematics are applied to patient-specific reconstructions of the bony anatomy; the femur rotates within the acetabulum until bone-bone contact is achieved, from which terminal range of motion (ROM) and impingement locations are approximated. A study by Kapron et al. (2014) found that predictions of impingement location in bone collision models often did not agree with locations of impingement when patientspecific kinematics and soft tissue surfaces were included in models. The generalized joint kinematics and assumption that impingement occurs where bones first contact disregards the importance of soft tissues, the actual structures being impinged, and how their morphology may affect impingement. Thus, predictions from bone collision techniques are likely inappropriate for their intended clinical applications for cam FAIS. In an effort to balance patient-specificity and computational simplicity, several research groups have developed, validated, and applied another streamlined modeling technique used to investigate soft tissue contact in articular cartilage of the knee (Hosseini et al., 2010; Li et al., 2005; Liu et al., 2010; Van de Velde et al., 2009; Yin et al., 2017) and ankle (Bischof et al., 2010; Wan et al., 2008, 2006). For clarity, we will refer to this modeling method as “soft tissue overlap” (STO) modeling. The STO modeling approach includes four primary steps: (1) Patient-specific 3D surface 8 meshes of bone and soft tissue are created from computed tomography (CT) or magnetic resonance (MR) images; (2) 3D bone reconstructions are then registered against images acquired using stereophotogrammetry techniques to extract patient kinematics in a motion of interest; (3) kinematics are then applied to the 3D surface meshes, such that bones move as rigid bodies with their contiguous soft tissue meshes; and (4) contact area and strain (Δ L / L ) are calculated where overlap between opposing soft tissue surfaces occurs. This method assumes large soft tissue deformation occurs where soft tissue surfaces overlap. The STO modeling method has never been applied to the hip joint, but it could have great potential, especially for the investigation of cam FAIS, given that abnormal bone anatomy is hypothesized to alter labral contact mechanics within the hip. With the high degree of variability in diagnostic measures, individual morphology, and symptom severity, the STO modeling method could provide insight into labral mechanics that may help researchers and clinicians better understand the condition itself. Because STO modeling does not require computational optimizations, it could be applied to provide fast solutions, such as preliminary investigations before FE or DEA setup, or in clinical diagnostics, treatment and planning, and intraoperative guidance. Objective The overarching aim of this study is to investigate contact mechanics of the acetabular labrum, femur, and femoral cartilage in cam FAIS using the STO modeling pipeline. We hypothesize that cam FAIS models will show greater contact (area, depth, and strain) during walking and pivoting activities. We also hypothesize that labral contact will be more concentrated in the anterior region of the joint in the cam FAIS cohort given that joint damage is typically seen anteriorly (Banerjee and Mclean, 2011; Ganz et al., 2003; Groh and Herrera, 2009; Leunig et al., 2003). We applied patient-specific kinematics from two walking activities (level and inclined) and two pivoting activities (internal and external) to reconstructed soft tissues of the hip. Three STO predictions (contact area, depth, and strain) of contact at the labrum were evaluated between cam and control participants in three anatomical regions (anterior, posterior, and superior). This study setup will allow us to achieve our objectives in determining how and where labral contact mechanics 9 may differ in joints with cam FAIS, which we hope will increase understanding of cam FAIS and provide a framework for a computational tool that may aid in treatment options and prevention. Methods Participant Recruitment Six patients with cam FAIS and 10 controls were recruited for this study, which was approved by the University of Utah Institutional Review Board (IRB #510523) (Table 1.1.). Diagnosis of cam FAIS was determined based on patient reported symptoms, positive clinical examinations (i.e., anterior impingement exam), and confirmation of cam morphology on radiographic images in the modified false profile, frog-leg lateral, and anteroposterior positions (Nötzli et al., 2002). Cam morphology was quantified using the alpha angle, where angles greater than 55° were considered abnormal. All patients were scheduled for surgery at the time they were recruited for this study. Control participants had no history of hip pain and were screened using an anterior-posterior radiograph prior to study inclusion (Atkins et al., 2017; Fiorentino et al., 2017, 2016a, 2016b), where an alpha angle measured to be greater than 55° resulted in exclusion from the control group. All patients and control participants were recreationally active and had no previous history of lower limb surgery, a body mass index (BMI) less than 30 kg/m2, a lateral center edge angle between 20° and 45° (to exclude individuals with hip dysplasia and pincer morphology), and no radiographic evidence of OA. For a visual description of the participant recruitment process, refer to Figure 1.4, panel [A]. Participant Imaging Computed Tomography Arthrograms CT images of the hip and distal femur were acquired with a SOMATOM Definition 128 CT scanner (Siemens AG, Munich, Germany). For the proximal femur and pelvis, images were acquired at 120 kVp, 1.0 mm slice thickness, and 200 to 400 mAs with variable fields of view due to participant size (Henak et al., 2014). For the distal femur, images were acquired at 120 kVP, 3.0 mm slice thickness, and 150 mAs. The distal femur was imaged with CT so that anatomical 10 coordinate systems could be defined when calculating hip kinematics (Harris et al., 2012; Kapron et al., 2014). Refer to Figure 1.4, panel [B]. Dual-Fluoroscopy Motion Capture Following a validated protocol (Kapron et al., 2014), participants performed several activities in a custom dual-fluoroscopy (DF) system (Radiological Imaging Services, Hamburg, PA). The DF set up consisted of two stationary x-ray emitters (Housing B-100/Tube A-142, Varian, Salt Lake City, UT) and two stationary 12” image intensifiers (T12964-P/S, Dunlee Inc., Aurora, IL), capturing video at 100 Hz during four activities of daily living: level walking, inclined walking, internal pivots, and external pivots. We chose these activities because they can induce impingementrelated symptoms in patients with FAIS (Clohisy et al., 2009). Two trials were obtained for all activities, with DF capturing the full gait cycle in both trials. Level walking occurred at the individual’s preferred speed (as determined by a timed walk) on an instrumented treadmill (Bertec Corporation, Columbus, OH, USA) set at a 0° incline. The treadmill was then raised 5° to evaluate inclined walking at self-selected speed. For the internal pivot, participants rotated their trunk ipsilaterally (toward the hip of interest), such that their leg rotated medially in relation to their pelvis, while their feet remained planted. The same procedure was followed for external rotation; however, participants rotated their truck contralaterally to the end of range of motion. Due to soft-tissue Xray scatter, the field of view for pivoting was configured such that the position of the pelvis and femur was evaluated only at the maximum rotated position. Refer to Figure 1.4, panel [C]. Model Construction and Kinematics Bone and Soft Tissue Models Original hip CT images were first resampled to three times the input size to improve resolution and facilitate accurate segmentation (Harris et al., 2012; Kapron et al., 2014). Image stacks were segmented semiautomatically in Amira (versions 5.6.1-6.2.0, Visage Imaging, San Diego, USA) to delineate the pelvis, femur, femoral cartilage, acetabular cartilage, and acetabular labrum (Henak et al., 2011). Surface segmentations were smoothed and decimated and then 11 converted into a volumetric mesh of tetrahedral elements automatically in Amira (Harris et al., 2012). This smoothing process reduced mesh densities to ~2.25 faces/mm 2 for bone and ~3 faces/mm2 for soft tissues. Refer to Figure 1.4, panels [D-E]. Model-Based Markerless Tracking Kinematic positions of the three-dimensional (3D) surface reconstructions of the pelvis and femur bone were calculated using model-based markerless tracking of the DF images (Bey et al., 2008, 2006). The entire gait cycle (heel-strike to heel-strike) was tracked for walking and inclined walking. However, for internal and external pivoting, the hip was not visible in the field of view of the DF system for the entire trial. Therefore, we tracked 20 frames prior to and following the end range of motion for internal and external pivoting, since both the femur and pelvis bones were visible across these frames. Model-based markerless tracking has been validated in the hip by Kapron et al. (2014) who reported maximum kinematic errors to be 0.48 mm in translations and 0.6º in rotations. The trial that provided the clearest dual-fluoroscopy images for (or greatest range of motion for the pivoting trials) was selected for analysis. Refer to Figure 1.4, panel [F]. Filtering and Interpolation of Kinematics To reduce noise as a result of small deviations in tracking, all kinematics were smoothed with a fourth-order bidirectional low-pass Butterworth filter with a cut-off frequency of 10 Hz. After filtering, the walking trials were normalized between participants by linearly interpolating the bone trajectories such that 0% gait corresponding to the first heel-strike event, 60% to the following toeoff event, and 100% to the second heel-strike event (Bergmann et al., 2001). For the pivoting trials, 20 frames of data (0.2 seconds) were tracked pre- and postterminus position across all participants. With these extra frames, the Butterworth filtering could still be applied to the pivoting activities, to help smooth potential noise at the event of maximal range of motion. Filtering and interpolation calculations were all performed in MATLAB (v7.10; The MathWorks, Natick, MA). 12 Contact Analysis Anatomical Orientation and Division To ensure that anatomical divisions were consistent across all participants, all meshes were transformed from their original CT positions to anatomical orientations established by Wu et al. (2002) and recommended by the International Society of Biomechanics wherein the origin was defined as the femoral head center (see Table 1.2 for axis definitions). All transformations of the 3D reconstructions and associated kinematics were performed in MATLAB. The labral meshes were divided into anterior, posterior, and superior anatomic regions. To ensure these divisions remained consistent across participants, each mesh was projected onto a plane defined by bony landmarks of the pelvis, namely (1) the acetabular rim of the pelvis; and (2) the lunate center. The 2nd principal curvature tool and sphere fit tool in PostView (FEBio Software Suite v2.4.0, University of Utah, Salt Lake City) automated the landmark identification process, which provided reproducible definitions across the participants (Uemura et al., 2019; Atkins et al., 2018; Kapron et al., 2014). Using singular value decomposition of the acetabular rim points selected by PostView, a plane was created, with the plane’s origin defined by the normal projection of that plane to the lunate center. All calculations were performed in MATLAB, using code developed for previous studies (Atkins et al., 2018; Uemura et al., 2019). Chondrolabral Boundary Definition Computed tomography images do not consistently provide a clear boundary between the acetabular cartilage and labrum (chondrolabral boundary) through typical quantitative evaluation of voxel intensities in segmentation software. Therefore, the boundaries were defined manually in PreView (FEBio Software Suite v2.2.0, University of Utah, Salt Lake City) by referencing anatomical cues (such as the acetabular rim, the curvature of the cartilage and soft labrum surface, and the sulcus) as has been performed for previous studies (Henak et al., 2011; Todd et al., 2018). 13 STO Calculations of Area, Depth, and Strain Kinematics derived from model-based tracking were applied to the femoral cartilage (following femoral kinematics) and labrum meshes (following pelvic kinematics) using the Kinemat tool in PostView (FEBio Software Suite v. 2.0-2.2, University of Utah, Salt Lake City) (refer to Figure 1.4, panel [G]). Then, using the distance map tool in PostView, distances between the femoral cartilage and labrum were calculated (d0 in Figure 1.5). This distance, d0, was reported as the overlap depth value. Any area of overlap between the two surfaces was calculated as a negative distance value by the distance map tool. By thresholding for only negative distance values, the area of overlap was calculated using the “Measure Area” tool in PostView. To control for differences in geometries (such as labral height from base to apex) between participants, the calculated area was divided by the total area of each anatomical region (anterior, posterior, superior, and overall) and multiplied by 100 to give a percentage of the area in contact. To calculate a strain estimate, the overlap depth value was divided by the sum of the labral and femoral cartilage thicknesses (Bingham et al., 2008; Bischof et al., 2010; Englander et al., 2018; Li et al., 2005; Yin et al., 2017), along the same normal vector projection. In other words, all distances (d0, df, and dL) were calculated in reference to the surface normals of the labral mesh, which ensured that all strain predictions were calculated along the same direction for each face of the mesh. In regions of high curvature, normal vector projections would extend infinitely, or would project onto some other distant portion of the mesh topology not relevant to the articular region of the joint. To circumvent this issue, faces from the apex of the labrum and on the back facing side of the labrum (defined using the 2nd principal curvature tool in PostView) were not included in distance computations. Statistical Analysis All statistical tests of the data were performed in MATLAB. Two-tailed, unpaired, weighted Student’s t-tests (alpha value = 0.05) compared area, depth, and strain between the cam and control participants across four anatomical divisions (total, anterior, posterior, and superior labrum) and four activities, with the three predictions of contact area, depth, and strain. For the pivoting 14 activities, only the end of range of motion event was considered for the area, depth, and strain calculations. The walking activities, however, compared the max and mean of each of the three contact predictions. As a separate statistical analysis, we employed statistical parametric mapping (SPM) to account for multiplicity issues in the time-domain dependence of contact area, depth, and strain to the normalized gait-cycle. SPM utilizes random field theory to compute cluster-based p-values and generates a threshold, the t* static, to define significant differences between datasets that are inherently similar in regard to their geometry and activities in time (Pataky et al., 2015; Vanrenterghem et al., 2012; Warmenhoven et al., 2018). SPM has been used in other biomechanical contexts to analyze human movement, because of the inherently strong spatiotemporal connections in biomechanical problems (Baumgart et al., 2017; Fox et al., 2017; Schuermans et al., 2017; Vanrenterghem et al., 2012; Whyte et al., 2018). A MATLAB-based opensource software was used to perform all SPM calculations (www.spm1d.org, 1d version 0.4). Results Internal and External Pivots Student’s t-test results did not indicate significant differences between cam and control participants in the pivoting activities for any of the three calculations (area, depth, and strain) in the labrum overall as well as in any of the anatomical divisions of the labrum (Figure 1.6). However, in general, external pivots showed more pronounced differences between cam FAIS and control participants than did the internal pivots. Level and Inclined Walking For the level-walking activity, Student’s t-tests of the averaged and mean and maximum contact area, depth, and strain did not indicate significant differences between cam and controls. This was true for the labrum overall and in the three anatomical divisions of the models (anterior, posterior, and superior) (Figure 1.7, top). However, the inclined-walking trials showed that the cam cohort had significantly greater mean contact area and strain, as well as maximum strains in the 15 anterior labrum (Figure 1.7, bottom). Of the comparisons that showed significant differences, maximal strains of the anterior labrum for the inclined walk showed the largest group-wise differences (Figure 1.7, bottom). T-tests performed across the level-walking trials found significantly higher contact area and strain in the cam group in the posterior labrum directly following heel-strike (4-9% and 3-9% of the gait cycle for contact area and strain, respectively) (Figure 1.8, third row, first and last columns). Contact area was also slightly but significantly greater for the cam cohort in the anterior labrum in 93-94% of the gait-cycle, whereas the other two predictions, depth and strain, showed no significant differences for the entirety of the trial (Figure 1.8, second row, first column). T-tests performed across the inclined-walking trials found that the cam cohort had significantly greater contact area during 25-35%, 69-74% of the gait cycle, and greater strain during 25-35%, 70-75% of the gait cycle in the anterior labrum (approximately during the swing and midstance phases of walking) (Figure 1.9, second row, first and last columns). T-tests also showed significantly greater contact area and overlap depth in the superior labrum of the cam FAIS cohort during 52-57% and 1-2% of the gait-cycle, respectively (Figure 1.9, third row, first and second columns). Upon correcting for multiple comparisons with SPM, no significant differences were found between the predictions from the STO model for any of the activities (Figure 1.10), which was true for all comparisons, even those in which significant differences were found between the cohorts using Student’s t-tests at every point in the gait cycle (indicated with asterisks in Figures 1.8-1.9). Discussion In this study, we analyzed labral contact mechanics of the acetabular labrum using a soft tissue overlap (STO) modeling method (Bingham et al., 2008; Bischof et al., 2010; Englander et al., 2018; Li et al., 2005; Yin et al., 2017) to investigate if and labral contact may be altered by cam FAIS. Clinical observations led us to hypothesize that hips with cam FAIS have increased labral contact, primarily in the anterior region of the acetabular labrum during dynamic activities such as walking and pivoting (Clohisy et al., 2009; Griffin et al., 2016; Groh and Herrera, 2009; Leunig et 16 al., 2003). Overall, our results did not conclusively show differences in articular contact between the two groups, but several interesting trends did arise that could inform future applications of STO methodologies for future studies. During preliminary diagnostics for cam FAIS, clinicians test the end range of motion during passively applied exams that are thought to induce impingement (Griffin et al., 2016). Several of the most common exercises include internal (impingement exam) or external (FABER exam) rotations of the femur, in combination with flexion (and abduction in the FABER test) to exacerbate impingement (Griffin et al., 2016; Kapron et al., 2012). Therefore, we hypothesized that weightbearing pivoting exercises may also induce greater labral contact or impingement in the cam FAIS participants. Although t-test results did not find any significant differences between the cam FAIS and control participants, we found noteworthy differences between the two activities that could guide future studies. The mean and variance of each cohort was nearly identical for the internal pivots with all three STO predictions and the different anatomical divisions (Figure 1.6, first row), whereas external pivots showed greater disparity between the two cohorts, with cam FAIS participants showing greater contact area, depth, and strain in all anatomical divisions of the joint. These trends may indicate that external pivots would be valuable to study again. For example, our sample size was small, as was the number of trials collected per participant, so perhaps by increasing both, variance would decrease, and significant differences would become apparent. It may also be useful to investigate clinical diagnostic exams with increased external rotation, as is done in the FABER test, used as a diagnostic tool to induce impingement-related pain in patients with cam FAIS (Kapron et al., 2014). Walking trials have been reported to induce pain for about 60% of patients with cam FAIS (Clohisy et al., 2009), and inclined walking tends to increase flexion angles (Bergmann et al., 2001). For this reason, we hypothesized that participants with cam FAIS would show greater anterior labral contact than controls, especially during inclined walking. Although trends in the inclined walk indicated that there could be significantly (according to traditional t-test results) greater contact area and strains in the anterior labrum of the cam FAIS cohort, the timing of those differences in the gait cycle did not occur when we hypothesized (Figure 1.8). Specifically, we anticipated greater contact 17 during times in the gait cycle where flexion was high; however, these differences were most pronounced during midstance and swing phases of gait, where flexion is low (Ganz et al., 2008; Griffin et al., 2016). The t-test results for level walking showed even more surprising trends; the cam cohort had greater posterior labrum contact area and strains directly after heel-strike (Figure 1.7), which does not coincide with damage mechanisms from the literature, which report that soft tissue damage predomiantely occurs in the anterosuperior region of the joint. Greater posterior contact in the cam cohort may suggest that cam FAIS alters contact in both the anterior and posterior labrum, but because the anterior labrum is thinner than other regions, it may be more suseptible to damage (Banerjee and Mclean, 2011; Clohisy et al., 2009; Groh and Herrera, 2009). As with the pivoting trials, we suspect that a larger sample size and multiple trials could help us better determine whether these trends are consistent across cam FAIS patients. Future work could also benefit from investigating activities that have higher impact, or greater extremes in range of motion such as squats or lunges, running, jumping, or push-off activities. Investigating other activities with larger cohorts may help us to assess the meaningfulness of the trends we observed, but we recognize that certain limitations inheret to our model inputs could affect results. The first limitation that could have some influence on STO model predictive capability is the definition of the chondrolabral boundary. Herein, this boundary was defined manually using anatomical cues from the 3D surface reconstructions, which followed the process used for patient-specific FE models of the hip (Henak et al., 2011; Todd et al., 2018). Henak et al. (2011) reported substantial changes in FE predictions of labral load with alterations to the chondrolabral boundary. Our STO model predictions could also be largely affected by this boundary, as the labral region is essential to calculations of strain and contact area. Another limitation to the predictive capabilities of this modeling technique could be the inherent errors in model-based markerless tracking. Model-based tracking has errors <0.5 mm in translation and <0.6º in rotation (Kapron et al., 2014). These errors are similar to those reported for tracking of knee and ankle motion (0.1-0.24 mm, and 0.3-0.6º) (Caputo et al., 2009; Li et al., 2008). Still, given the physical size and geometry of the labrum, it is possible that even submillimeter and subdegree errors could have a noteable affect on predictions made by STO models. These limitations provided 18 the motivation to compelte the studies detailed in Chapter 2, where we investigate chondrolabral and kinematic variations and resulting STO prediction sensitivities, as well as provide a comparison of contact area and strain to a FE model. In summary, our results indicate minor differences in labral contact mechanics among hips with cam FAIS when compared to control hips during walking activities. Of the four activities we studied, inclined walking showed the largest differences among groups, and thus, this activity should be considered for future studies. Of the three predictions that we used to compare contact (area, depth, and strain), it seems that the area and strain predictions aligned more often than did the basic depth calculation, suggesting that STO calculations may be more appropriate to view together, rather than any single prediction alone. Finally, the inherent errors arising from limitations in our model reconstruction and kinematics may influence our STO predictions, which provides motivation to pursue sensitiivty studies as part of Chapter 2. 19 Figure 1.1. Schematics of the human hip joint. Left: An anterior view of the whole hip joint, including the femur, pelvis, articular cartilages (femoral and acetabular), and the labrum. Middle: A lateral view of the acetabulum, with the pelvis (white), acetabular cartilage (yellow), and labrum (pink). Right: A coronal cross section of the acetabulum highlighting the triangular structure of the labrum, and the chondrolabral boundary (the interface between cartilage and labrum) to the acetabular rim. 20 Figure 1.2. Schematics of the hip from a superior view.Top left: Typical healthy bone morphology of the femur and acetabulum. Top right: Pincer FAIS morphology, where excessive/abnormal cortical bone growth, shown in red, is found on the acetabular rim. Bottom left: Cam FAIS morphology, where excess cortical bone is found on the femoral head/neck junction (termed the “cam lesion”). The cam lesion is often found antereosuperiorly. Bottom right: Combination FAIS exhibiting both pincer and cam morphologies. Image adapted from JointPain.MD. 21 Figure 1.3. Depiction of impingement. A coronal cross-section depicting impingement in the hip (right side). Impingement is thought to occur when the femur and femoral cartilage (blue) come into contact with the acetabular cartilage (yellow) and labrum (pink) due to the cam lesion, which gives the femoral head a less spherical rotation arc. The cam lesion is thought to increase load as well as compressive and shearing stresses experienced in the labrum. 22 Figure 1.4. Methods flowchart. [A] Radiographs measures separated participants into cam (n=6) or control (n=10). Participants were imaged using CT arthrograms to obtain geometry [B], and dual fluoroscopy (DF) to obtain motions [C]. Model construction and kinematics [D-F] consisted of segmentation of the articular soft tissues [D], bones [E]. The 3D bone models were coupled with the DF videos to track motion [F]. Using all soft tissue models and bones, along with the kinematics found in tracking [G], we then could calculate the three STO predictions of interest to us, namely: area, depth, and strain [H]. Panel H is described further in Figure 1.5. 23 [H] Pelvis Bone Labrum Acetabular Cartilage dL Femoral Cartilage Overlap Region d0 df Strain Equation Femoral Head ɛ= 𝑑𝑜 𝑑𝑓 + 𝑑𝐿 Figure 1.5. STO model methodological schematic. A close-up view of panel [H] from Figure 1.4, depicting the region of the STO modeling. A cross-sectional view of overlap between the acetabular labrum and femoral cartilage is given with the strain estimate equation. The variable d o refers to the depth of overlap between the femoral cartilage and labral surfaces. This value was compared across participants and reported as the “depth” measure. The variable d f is the femoral cartilage thickness, calculated with the normal vector projection from the acetabular articular surface. The labral thickness, dL, was calculated from the same projection, but spanned the distance between the articular to nonarticular side of the labrum. 24 Figure 1.6. Internal/External pivots at end stance, or maximum range of motion (ROM). Area (left), depth (middle), and strain (right) were calculated across the entire labrum (all) and in the anterior, posterior, and superior labrum. T-tests found no significant differences between cam and control cohorts in any of the anatomical regions for either activity. 25 Figure 1.7. Mean and max walking trial results. Student’s t-tests were applied to compare cam and control cohorts in each dataset per anatomical division (overall, anterior, posterior, and superior) for overlap area, depth, and strain. No significant differences were found overall, or in any of the individual regions of the labrum during level-walking (top panel). During inclined-walking, cam participants had significantly greater mean contact area and strain and greater maximum strain between the two groups in the anterior labrum (noted with *); no other differences were found during incline walking (bottom panel). 26 Figure 1.8. Dynamic level walking results. Cam (red) and control (blue dotted) participants were compared with Student’s t-tests performed at each frame of the gait cycle. Comparisons were made overall and within individual anatomic regions (rows) for the area, depth, and strain (columns). The black dotted lines with asterisks (*) above notate significant differences p<0.05. The largest differences between the two cohorts were found in the posterior region (third row) when comparing area and strain during heel-strike (loading phase of gait). However, upon correcting for multiple comparisons with SPM, these differences were no longer significant (see Figure 1.9). 27 Figure 1.9. Dynamic inclined walking results. Cam (red) and control (blue dotted) participants were compared with Student’s t-tests performed at each frame of the gait cycle. Comparisons were made overall and within individual anatomic regions (rows) for area, depth, and strain (columns). The black dotted lines with asterisks (*) above notate significant differences p<0.05. The largest group differences were found in the anterior region (second row) when comparing area and strain during the swing and midstance phase of gait. However, SPM results showed that these differences were no longer significant (see Figure 1.9). 28 Figure 1.10. Selected comparisons of t-test and SPM results. SPM results from strain and area predictions in the posterior labrum during level-walking (top) and anterior labrum during inclined walking trials (bottom) are displayed. P-values from t-test results plotted above SPM{t} results. The SPM results can be seen to be nearing their critical t* thresholds (red dotted lines) when the pvalues are lower than 0.05 (red dotted line) in these regions, but none crossed the require t* threshold to determine significance. 29 Table 1.1. Participant demographics: Male/female (M/F) divisions, side of hip that was imaged (L/R), age, height, weight, and BMI. Demographics Sex (M/F) Side (L/R) Age (years) Height (cm) Weight (kg) BMI (kg/m2) All Participants 10/6 9/7 25 ± 5 175.8 ± 10.0 68.4 ± 11.7 22.0 ± 2.3 Controls Cam FAIS 6/4 4/2 5/5 4/2 23 ± 2 28 ± 7 174.6 ± 10.0 177.9 ± 10.5 64.7 ± 11.1 74.5 ± 10.9 21.1 ± 2.0 23.5± 2.2 Table 1.2. ISB standard for hip coordinate system, with axes definitions on the left column and rotational definitions on the right column. x y z + anterior/ - posterior + superior/ - inferior + lateral/ - medial rx ry rz + adduction / - abduction + internal / - external + fllexion/ - extension CHAPTER 2 SENSITIVITY STUDIES OF SOFT TISSUE OVERLAP MODELING AND COMPARISON TO FINITE ELEMENT MODELING Introduction, Motivation, and Objectives In Chapter 1, we applied soft tissue overlap (STO) modeling to investigate cam FAIS, a painful hip condition marked by abnormal anatomy thought to alter typical labral contact mechanics of the hip. Our STO models were created following the methodological pipelines established by others studying cartilage mechanics in the knee (Hosseini et al., 2010; Li et al., 2005; Liu et al., 2010; Van de Velde et al., 2009; Yin et al., 2017) and ankle (Bischof et al., 2010; Wan et al., 2008, 2006). Validation studies of STO models have reported that STO predictions of contact area differed from measurements in cadavers by ~14% in the knee (Bingham et al., 2008) and ~4% in the ankle (Wan et al., 2006). Differences between STO strain predictions and cadaver measurements have only been reported for the ankle, with a difference of ~6% (Bischof et al., 2010). STO has not been validated for analysis of hip contact mechanics. However, we conjecture that contact area and strain measures would be similar in the hip because reported errors for their validated kinematics and 3D mesh reconstruction techniques (max reported kinematic errors in the ankle and knee, respectively: 0.1 mm, 0.3º) are similar to errors in the hip (0.5 mm and 0.6°) (Bingham et al., 2008; Bischof et al., 2010; Kapron et al., 2014). Still, it is unknown to what extent the geometry and scale of the hip may affect STO predictions of labral mechanics. By investigating model sensitivities through systematic alterations of the parameters, we hope to better identify model limitations and provide suggestions for improvement in the future. The two major limitations that could have a notable impact on STO model credibility and reliability are: (1) the meshing process for the labrum, involving a manual definition of the 31 chondrolabral boundary (the border separating the labrum from acetabular cartilage); and (2) kinematic errors. To complement sensitivity studies, we compared STO model predictions to those provided by a finite element model that was developed using a validated pipeline and included the same participant-specific mesh and kinematic data. The chondrolabral boundary definition is a potential source of modeling error because it could alter the size and form of the labrum, which would in turn affect predictions, such as area of the labrum in contact. After establishing the manual chondrolabral boundary selection process that was used in our study, Henak et al. (2011) found that shifting the chondrolabral boundary in their FE model substantially altered predicted load support by the labrum. Thus, we theorized that predictions from the STO model would also be affected by a shift in the chondrolabral boundary. More specifically, we hypothesized that under-selection of the labral region would result in lowered depth, area, and strain predictions, whereas over-selection would result in higher predictions. Kinematic errors were identified as a second potential source of modeling error because the kinematic error of dual-fluoroscopy (0.5 mm) (Kapron et al., 2014) was not insignificant when compared to the observed overlap depth, which was typically between 1-2 mm, with maximal values reaching 3.5 mm (Figures 1.6-1.9). We hypothesized that STO predictions, especially depth and strain, would be sensitive to kinematic variations, even when the region of maximal overlap and strain were not along the same axis as the kinematic deviation. Finally, we were interested in understanding how STO predictions compared to a finite element model with the same participant-specific 3D reconstructions and kinematics. FE has been an invaluable computational modeling tool, allowing researchers to explore the mechanical behavior of the hip that would otherwise be extremely difficult or impossible to perform in vivo or in cadaver studies. In fact, no in vivo labral mechanics studies exist in which to compare STO predictions; existing animal and cadaveric studies report deformation measures and strain definitions that are not directly comparable to STO predictions (Ollivier et al., 2017; Safran et al., 2011). Although many assumptions must be made in FE models of the hip because of physical uncertainties and parameter optimization constraints, we consider FE to be a reference standard modeling tool for comparison because FE models of the hip provide accurate predictions of hip 32 contact mechanics (Anderson et al., 2008; Henak et al., 2014). With our investigation of sensitivities to the chondrolabral boundary and kinematic errors, coupled with a comparison of STO and FE modeling, we aim to highlight the advantages and weaknesses of STO and its use in the hip. Findings from this chapter should allow researchers to continue developing mechanical modeling systems to address the needs of cam FAIS treatment and diagnosis. Methods Chondrolabral Boundary Sensitivity All participant models that were investigated from the previous study in Chapter 1 were included for analysis of the sensitivity of STO predictions to alterations in the chondrolabral boundary (refer to Table 1.1). To understand the sensitivity of the STO predictions to manual placement of the border, the chondrolabral boundary was shifted either toward the labral apex (shrinking the defined region of the labrum) or away (growing the defined labral region), and the resulting changes in STO predictions were compared. The chondrolabral boundary was defined by nodes on the surface meshes, which connected mesh elements from both the labral region and the acetabular cartilage (Figure 2.1, left). New borders were created by defining a radius, r, away from each of the boundary nodes, and selecting any surface nodes that fell within this radius (Figure 2.1, middle). The selected nodes could then be either added to or removed from STO model calculations, effectively enlarging or shrinking the defined area for the labrum. A medial shift of the border enlarged the overall labral region, whereas a lateral shift shrunk the labral region (Figure 2.2). Because of the constraints of our mesh density (~3 elements per mm2), the radius r was shifted by increments of 0.25 mm (up to ±10 mm). We chose to investigate how changes to the chondrolabral boundary altered STO model predictions for inclined walking because this activity demonstrated larger differences in labral mechanics between patients with FAIS and control participants. After varying the labral region with the chondrolabral boundary shift, the contact area, maximal depth, and strain values were recalculated for each inclined walking activity. The mean of each STO prediction (area, max depth, 33 and max strain) was calculated after the boundary shift, 𝑋𝑟 in Eq. (2.1), divided by the mean of the STO predictions with their original chondrolabral boundary, 𝑋0 in Eq (2.1), which was used to find the percent change as follows: % 𝐶ℎ𝑎𝑛𝑔𝑒 = (1 − 𝑋𝑟 𝑋0 ) × 100 % (2.1) The percent change was averaged across participants for each increment in r up to ±10mm, and then plotted with 95% confidence intervals to show general trends and variations in each STO prediction. All calculations and statistics were performed in MATLAB. Kinematic Sensitivity We selected four participant models to investigate the sensitivity of STO predictions to changes in applied hip kinematics (two male/female, two left/right hip sides, and two cam/control). Translational and rotational deviations were introduced to our kinematic solutions along the three primary anatomical axes defined for the hip anatomical coordinate system (refer to Table 1.2 in Chapter 1). Translational deviations varied between ±1 mm and rotational deviations between ±1º in increments of 0.2 mm and 0.2º, respectively. Linear interpolation was used to estimate the STO predictions at 0.5 mm, the maximum reported translational errors of our dual-fluoroscopy system. All translational and rotational deviations were calculated in MATLAB. The same methods used to calculate contact area, depth, and strain (outlined in Chapter 1, Methods) were then applied to the varied datasets. Sensitivities of area, depth, and strain predictions were found by comparing the mean of the altered kinematic solutions against the mean of the original STO solutions, as was done for the chondrolabral boundary sensitivity study (Eq. 2.1). Finite Element Model Comparison For the final portion of our study, we compared contact area and strain predictions from an STO model to a single FE model of a control participant (female, 20 years, 20.3 kg/m2). The FE model meshes of soft tissues were made of volumetric linear tetrahedral elements (femoral 34 cartilage, acetabular cartilage, and labrum) and triangular surface meshes for bones (femur and pelvis). Bones were modeled as rigid bodies. The constitutive equations governing the material behavior of the labrum and cartilage followed Mooney-Rivlin models (Table 2.1). Cartilage is highly hydrated and is often modeled with biphasic materials to capture the fluid behavior of the tissue. However, the activities we investigated in Chapter 1 have a higher loading rate (e.g., one gait cycle is typically less than 1 second in length). During higher loading rates, fluid does not have sufficient time to exude from the solid matrix, and thus, cartilage can be assumed to act as a hyper-elastic, nearly incompressible solid (Abraham et al., 2013; Anderson et al., 2008; Knight et al., 2017; Todd et al., 2018). Modeling bone with a rigid material approximation effectively increases cartilage contact stresses calculations and decreases contact area on the cartilage (Anderson et al., 2010, 2008). The influence of the rigid bone assumption on predictions of labral mechanics has, to our knowledge, not been evaluated. Regardless, the STO approach assumes rigid bones, and thus, to enable direct comparisons, we assumed bones acted as rigid bodies in the FE model (Table 2.1). Contact definitions between material interfaces are provided in Table 2.2. We chose to model the terminus position of an internal rotation activity in our FE model because this activity showed a greater amount of contact strain and contact area than did the external pivot for this participant. Because the STO model used a strain calculation that was defined according to the surface normals of the labral mesh, we compared this to 1st principal Lagrangian strain in FEBio (also derived from the surface normals). We also compared STO model predictions to effective Lagranian strain, which can be equated to von Mises stresses, used as criteria for material failure. Results Chondrolabral Boundary Sensitivity As the chondrolabral boundary was shifted laterally (shrinking the defined labral area), all three STO predictions of area, depth, and strain decreased compared to their original values. At approximately -10 mm, contact area, depth, and strain reached -100% change, indicating all three predictions reached the value of zero, meaning that the labral area also reached zero across all 35 participants (Figure 2.3). When the boundary shifted medially (increasing labral size), depth and strain estimates increased by approximately 1-3% per mm. Shifting the boundary laterally (shrinking labral size), decreased depth and strain estimates approximately 4-12% per mm (Figures 2.3-2.4). Strain values plateaued at +5 mm, with 26% overestimation from the original average strain value for inclined walking (Figure 2.3). Area remained at a 0-2% difference from the original calculations until approximately ±3 mm (Figure 2.3-2.4). Contact area was the least sensitive prediction to shifts in both directions compared to the depth and strain predictions, but it also showed the greatest amount of variance in the lateral boundary shifts. Overlap depth had the least amount of variance across both directions but was more sensitive to lateral boundary shifts after approximately +2 mm. Kinematic Sensitivity Area, depth, and strain predictions had relativity similar sensitivities in each of the different anatomical axes. The inferior/superior translations produced the greatest observable variance across all three predictions. Area was the least sensitive parameter to rotational errors, ranging between 0.04-0.7% change along the three anatomical axes at maximum reported kinematic errors (0.6°), whereas overlap depth and strain ranged from ±0.6-6.5% across all axes. Posterior translations resulted in increased estimations across all predictions, whereas predictions decreased when translations occur anteriorly. Both inferior and superior shifts resulted in increased estimations of all predictions, except strain, which decreased with greater inferior translations. Medial translations resulted in only slight overestimations in area and depth, but with a decrease in strain. Lateral translations resulted in overestimation for all three predictions. See Table 2.3 for values at the maximum reported dual-fluoroscopy errors at ±0.5 mm and ±0.6° across all tests. Overall, the model was much more sensitive to translational errors (Figure 2.5) than rotational errors (Figure 2.6) about the three axes; translational errors in the inferior/superior plane showed the greatest variance. 36 Finite Element Model Comparison Qualitatively, the region of contact area estimated in STO and FE was similar, with the only major disparity between the models being that STO showed contact near the labral apex whereas the FE model did not (Figure 2.7). Contact strain showed an even more pronounced disparity at the labral apex, in that STO strains were more concentrated, whereas the FE model displayed more diffuse strains, with maximum strain closer to the chondrolabral boundary (Figure 2.8). At the terminus position of an internal pivot activity, the STO model predicted 30% contact area of the labrum, with the FE model predicting 23%. This 7% difference was the greatest disparity observed between models from their original unloaded configuration to terminal stance of an internal pivot trial (Figure 2.9). The STO model also overestimated contact strain as compared to the FE model (both effective and 1st principal Lagrange strains), estimating 67% strain, while the FE model estimated 64% and 55% strain for effective and 1st principal Lagrange strain, respectively (Figure 2.9). Discussion The goal of this chapter was to investigate the sensitivities of STO model predictions to modeling assumptions, mesh construction, and kinematic inputs. Specifically, we investigated the effect of variations in the chondrolabral boundary and kinematic solutions on STO predictions. We also compared STO predictions with those provided by an FE model that had equivalent mesh geometry. Overall, STO model predictions were sensitive to changes in the chondrolabral boundary, especially when the boundary shifted laterally, thereby shrinking the area of the labrum available for contact. Kinematics were most sensitive to translational errors along all anatomical axes and across all predictions; area was the least sensitive to variations in rotation. Finally, FE and STO models predicted similar regions of contact area and strain along the joint circumferentially, but both the contact area and strain estimates were larger in the STO model; the medial to lateral distribution of strain was more concentrated near the labral apex in the STO model. We chose to study the chondrolabral boundary because its definition affected the calculated area of the labrum, which was crucial to STO calculations. A previous FE study found 37 that load predictions in the labrum changed considerably (up to 4x in some cases) with an altered boundary definition (Henak et al., 2011). In general, we found that area was the least sensitive STO prediction to boundary shifts, and overlap depth showed the least amount of variance. Thus, area and overlap depth predicted by STO models may more reliable than strain predictions. Another interesting trend we found was that STO predictions were less sensitive to medial boundary shifts than lateral shifts, meaning that STO accuracy may be improved by erring on the side of labral over-selection rather than under-selection, especially when depth or strain predictions are of primary interest. Nevertheless, additional validation studies are needed to identify the true error associated with selection of the chondrolabral boundary. For example, cadavers could be scanned before and after the labrum has been dissected to evaluate the accuracy of the chondrolabral boundary. As a complement to the cadaver studies, inter- and intraobservational studies could help to determine the repeatability of the manual selection process and identify areas where this methodology could be improved. Other imaging modalities with higher spatial resolution, such as magnetic resonance imaging, may provide better visualization of the chondrolabral boundary. In Chapter 1, we found that overlap depth values typically ranged between 1-2 mm, and the maximum reported translational errors for our dual-fluoroscopy system reached 0.5 mm (Kapron et al., 2014). Similar errors have been reported for the knee and ankle (0.1-0.24 mm, and 0.3-0.6º) (Caputo et al., 2009; Li et al., 2008). Validation studies of STO models in these joints have reported STO model predictions and cadaver measurements of contact area differed by approximately 14% in the knee (Bingham et al., 2008) and 4% in the ankle (Wan et al., 2006); previously ~6% difference in strain between STO predictions and cadaver measurements has been reported when studying ankle biomechanics (Bischof et al., 2010). Our model was more sensitive to translational deviations than rotational deviations, a result that may not hold true for the knee and ankle joint, in which articular surfaces are approximately planar, whereas surfaces in the hip are approximately spherical. Consequently, we recommend that future sensitivity studies using STO modeling in other joints perform sensitivity studies to evaluate the influence of kinematics (with translational and rotational deviations. Although they are much easier to collect, kinematics derived from skin markers or other generalized kinematic data would likely not be appropriate for STO 38 modeling. Skin marker kinematics, although technically participant-specific, have reported maximum translational errors approximately 5x greater than dual-fluoroscopy. In general, the STO model overestimated strain and underestimated contact area when compared to the FE model, suggesting that our STO model prediction may not be adequate for studies requiring extreme accuracy, such as predicting the mechanical material failure of the labrum. However, STO modeling is likely adequate for comparative studies requiring more precision over accuracy, for example, in studies investigating pathological or surgical alterations to hip morphology. Circumferentially, the region of contact between the models agreed well; thus, STO modeling may be useful to determine where impingement occurs circumferentially. Additionally, the agreement in contact area in the models suggests that STO could be a useful tool in FE modeling and DEA mesh/spring optimization. FE and DEA models require finer mesh/spring resolution where contact occurs, which is not always known, meaning STO could help these models refine mesh/spring density only where contact is likely to occur to optimize computational time. Although we reported only one model and one contact position, the relative trends of the plots were similar as the models moved from an unloaded configuration to the internal pivot terminus position (Figure 2.9). If this trend were to continue in dynamic trials such as in walking, current disparities between the models could potentially be accounted for with the use of scaling factors. The most discerning difference between the FE and STO model solutions, however, may not be adjusted so easily. Circumferentially, the region of contact agreed between the models, but along the medial to lateral direction, strain and contact area were concentrated near the apex of the labrum in the STO model and the chondrolabral junction in the FE model. The differences between our STO and FE models near the labral apex draw attention to the unique geometry of the labrum and the fact that STO assumptions oversimplify the deformation patterns of this peripheral soft tissue structure. The simple STO calculation of strain (Eq. 2.1) may be more appropriate for cartilage or other soft tissues bounded by bone nearly parallel to the soft tissue surfaces (where loading could be equated to unbounded compression tests), but the labrum is bounded by bone only near the chondrolabral boundary, with the apex free to deflect in response to increased loads and strains. The labrum also exhibits anisotropic or transversely isotropic material behavior along the circumferential direction, 39 which we included in our FE model, possibly causing additional disparities between FE and STO (Henak et al., 2014; Petersen et al., 2003). Future studies may benefit from the addition of weighted factors, which scale STO strain according to the distance that surface overlap occurs from the labral apex to acetabular rim apex, to reduce errors that may result from STO assumptions. The benefits of computational complexity must be weighed against computational time to determine whether such actions would be of any real benefit to researchers or clinicians. Our sensitivity studies and STO to FE model comparison had several limitations. In our evaluation of the chondrolabral boundary, we did not address how errors could arise from our semiautomated segmentation. These errors are not known for the hip and would thus be a good additional parameter to investigate in future studies. A major limitation of the kinematic errors study is our sample size. Participants selected for this study were chosen in part because they showed a good range of labral contact throughout their trials. Additionally, we investigated the effects of kinematic error only on the labrum overall, not the divisions of the three anatomical divisions we evaluated in Chapter 1. By including participants who showed less contact overall and investigating contact in anatomical regions (anterior, superior, and posterior), we would predict even greater deviations in STO outcomes from kinematic error. Finally, the major limitation in our STO to FE comparison was that we had only one participant and one activity. Future studies should compare multiple subjects with several dynamic activities. In summary, although STO modeling has been validated in the ankle and knee joints, with reported errors that would likely be appropriate for comparison of pathologies in research or clinical settings (Bischof et al., 2010; Wan et al., 2006), further validation studies are required to determine the reliability of STO modeling for the hip. The hip may be prone to modeling assumption inaccuracies due to its unique geometry, which may or make STO more difficult to implement in the hip than in other joints. Validation studies investigating model construction and kinematics could likely improve STO outcomes and help investigators weigh the benefits of the STO method against its costs to predictive accuracy. We also suggest that sensitivity studies investigating model construction errors or kinematic errors be conducted for other joints to understand more fully the reliability of STO results and determine its proper use in research studies or in clinical endeavors. 40 Figure 2.1. Part I of chondrolabral boundary computational methods. Images depicting the method by which nodes were selected to be added or removed from labral calculations from the chondrolabral boundary (boundary nodes in yellow in right image). Lateral Labrum S P A I Cartilage -r +r Original Boundary Lateral Boundary Medial Medial Boundary Figure 2.2. Part II of chondrolabral boundary computational methods. Images depicting shifts in the chondrolabral boundary definition either medially or laterally by some distance r, thereby growing (+r, medial boundary) or shrinking (-r, lateral boundary) the region defined as the labrum (pink) in STO calculations. 41 Figure 2.3. Chondrolabral boundary sensitivity results (±10mm).The chondrolabral boundary was shifted up to 10 mm medially and laterally. The observed percent change in our averaged results of inclined walking trials was plotted against the boundary shifts in increments of 0.25 mm. Figure 2.4. Chondrolabral boundary sensitivity results (±4mm). Confidence intervals (95%) included in the plots to highlight variance in the data. 42 Figure 2.5. Translational error results in kinematic sensitivity analysis. Average percent change in STO calculations of area, depth, and strain (top, middle, and bottom rows) after the introduction of translational errors along the three anatomical axes (columns) to inclined walking activities across four participants. Plotted in increments of 0.1 mm up to ± 1 mm. Confidence intervals (95%) were included to highlight data variance. Dotted lines along ± 0.5 mm are included to compare the plots to maximum reported errors for our validated kinematics. 43 Figure 2.6. Rotational error results in kinematic sensitivity analysis. Average percent change in STO calculations of area, depth, and strain (top, middle, and bottom rows) after the introduction of rotational errors along the three anatomical axes (columns) to inclined walking activities across four participants. Plotted in increments of 0.1° up to ± 1°. Confidence intervals (95%) were included to highlight data variance. Dotted lines along ± 0.6° are included to compare the plots to maximum reported errors for our validated kinematics. 44 Figure 2.7. Qualitative comparison of contact area, STO vs. FE modeling. Comparison of contact area predictions from STO (left), and FE (right) at the terminus position of an internal pivot. View 1: the acetabulum from a lateral position. View 2: superior portion of the acetabulum. Labrum and acetabular cartilage are both displayed, the labrum being superior to the cartilage, with the structures separated by a thin black line. Circumferentially, the contact regions were similar, but the STO model showed more contact near the labral apex than FE. 45 Figure 2.8. Qualitative comparison of contact strains, STO vs. FE modeling. Comparison of strain predictions from STO (left) and FE (middle/right: effective strain/ 1st principal Lagrange strain) at the terminus position of an internal pivot. View 1: the acetabulum from a lateral position. View 2: superior portion of the acetabulum. Labrum and acetabular cartilage are both displayed, the labrum being superior to the cartilage, with the structures separated by a thin black line. Circumferentially, the region of strain was similar between STO and FE, but the STO strain was less disperse and more concentrated near the apex of the labrum. 46 Figure 2.9. Quantitative comparison of contact area and strains, STO vs. FE modeling. Plots comparing the STO and FE predictions of contact area (left) and strain (right) as the models moved into the internal rotation terminus position from their original unloaded configurations. Contact area was ~7% greater in the STO model at the terminus position, the largest difference seen between the two models. At terminus position, predictions of strain were STO = 67%, FE (effective Lagrange strain) = 64%, and FE (1st principal Lagrange strain) = 55%. 47 Table 2.1. Material definitions and parameters of FE model. Mesh Articular Cartilages (Femoral and Acetabular) Labrum Bone (Femur and Pelvis) Material Definition Isotropic Mooney-Rivlin (uncoupled elastic) Anisotropic Mooney-Rivlin, Fibers running circumferentially (uncoupled elastic) Rigid Body Material Parameters density=0.854, c1=6.817, c2=0, K=1359 density=1, c1=1.4, c2=0, c3=0.05, c4=36, c5=66, =1.103, K=1000 N/A Table 2.2. Contact definitions of FE model. Contacting Surfaces Femoral cartilage Femur Acetabular cartilage Pelvis Labrum Pelvis Acetabular cartilage Labrum Femoral cartilage and labrum Acetabular cartilage and labrum Contact Definitions Rigid contact. Cartilage surface moves rigidly with the femur. Rigid contact. Cartilage surface moves rigidly with the pelvis. Rigid contact. Labrum surface moves rigidly with the pelvis No explicit definition required. The two structures share a boundary Frictionless, sliding contact Table 2.3. Kinematic sensitivity results. Average percent change in area, depth, and strain measures from the original averaged incline walk at the maximum reported translational errors (0.5 mm) and rotational errors (0.6º) for our validated DF system. Translations Area Depth Strain Posterior 9.4 15 2.3 Anterior -0.3 -4.2 -15 Inferior Superior 10 22 12 Medial 17 1.5 -19 Lateral Rotations Area Depth Strain Abduction -0.2 -0.6 -3.0 Adduction -0.5 1.6 3.4 External -0.7 -1 3.9 Internal 0.1 1.9 4.4 Extension 0.04 2.0 6.1 -8.1 0.3 -12 21 12 -2.9 Flexion -0.6 2.1 6.5 CHAPTER 3 SUMMARY, CONCLUSIONS, AND IMPACT The principal goals of this thesis were twofold: (1) to investigate labral contact mechanics in cam FAIS patients using the soft tissue overlap (STO) modeling method; and (2) to assess sensitivities of STO modeling in the hip to understand the reliability of results and to highlight important considerations regarding the use of STO modeling for future studies. In Chapter 1, despite not finding statistically significant differences between our cam FAIS and control cohorts, we did identify interesting trends that suggested cam FAIS may increase contact of the labrum. High variability in joint morphology and severity of impingement-related symptoms in cam FAIS patients may partially explain the lack of significant differences. Future work using STO modeling or other computational methods should consider including larger sample sizes and multiple trials per activity. In Chapter 2, the results of sensitivity studies indicated that model construction and kinematic errors could have a substantial influence on STO predictions. In particular, STO models were highly sensitive to translational kinematic errors, which suggests that the quality of kinematics used in STO models is a key determinant of STO predictive capabilities. For example, although much easier to collect, patient-specific kinematics derived from skin marker data are prone to substantial translational errors that would likely not be appropriate for STO modeling. Model-based markerless tracking of dual-fluoroscopy images, the method used to obtain patient-specific kinematics in this study, is currently the most accurate approach to measure in vivo joint kinematics, and therefore we recommend continued application of dual-fluoroscopy kinematics to drive STO models. However, additional validation studies are needed to determine the accuracy of STO for use in research and clinical studies. 49 The results of Chapter 2 demonstrated that the anatomical region of contact area and strain predictions of STO were consistent with FE, but the contact area and strain predictions were greater in the STO model. Differences between STO and FE models were especially apparent when comparing the location of peak contact strain, with peak strain located near the labral apex in the STO model, and near the chondrolabral boundary in the FE model. The described differences between the STO and FE models were most likely influenced by the STO rigid surface simplification, which did not allow for labral deflection as did the FE model. Additional scaling factors or other parameters may be necessary to account for the unique geometry of the labrum, which would deflect with loads in vivo. In cases where the objective of STO modeling is to compare pathologies or activities, such modifications may not be necessary. For example, if the goal of this modeling technique is use in clinical diagnosis or treatment (such as planning of surgical resections for patients with cam FAIS), the magnitudes of contact area and strain may be biased, but presumably this level of bias will be consistent across participants, making STO an option for comparative analyses. In summary, we found that there may be slight differences in labral contact patterns between cam FAIS and healthy populations using STO, but further studies are necessary to determine whether these findings can be generalized for cam FAIS. Future studies would benefit from larger sample sizes, activities with larger ranges of motion and higher joint impact, and multiple trials per activity. Our sensitivity studies found that STO was most sensitive to reduction of the labral size, and to translational deviations in kinematics, especially in the superior/inferior plane. From our comparison to an FE model, we also found that STO modeling assumptions skew the distribution of strain predictions near the apex of the labrum because STO models do not account for labral deflection. The findings from our sensitivity studies and FE model comparison emphasize the need for continued validation and sensitivity studies in the hip to assess the shortcomings and advantages of STO modeling, especially for its use in evaluation of highly variable conditions such as the evaluation of cam FAIS in clinical practice. REFERENCES Abraham, C.L., Maas, S.A., Weiss, J.A., Ellis, B.J., Peters, C.L., Anderson, A.E., 2013. A new discrete element analysis method for predicting hip joint contact stresses. J. Biomech. 46, 1121. Allen, B.C., Peters, C.L., Brown, N.A.T., Anderson, A.E., 2010. Acetabular cartilage thickness: accuracy of three-dimensional reconstructions from multidetector CT arthrograms in a cadaver study. Radiology 255, 544–52. Alzaharani, A., Bali, K., Gudena, R., Railton, P., Ponjevic, D., Matyas, J.R., Powell, J.N., 2014. The innervation of the human acetabular labrum and hip joint: an anatomic study. BMC Musculoskelet. Disord. 15, 41. Anderson, A.E., Ellis, B.J., Maas, S.A., Peters, C.L., Weiss, J.A., 2008. Validation of finite element predictions of cartilage contact pressure in the human hip joint. J. Biomech. Eng. 130, 051008. Anderson, A.E., Ellis, B.J., Maas, S.A., Weiss, J.A., 2010. Effects of idealized joint geometry on finite element predictions of cartilage contact stresses in the hip. J. Biomech. 43, 1351–7. Anderson, L.A., Anderson, M.B., Kapron, A., Aoki, S.K., Erickson, J.A., Chrastil, J., Grijalva, R., Peters, C., 2016. The 2015 Frank Stinchfield Award: Radiographic abnormalities common in senior athletes with well-functioning hips but not associated with osteoarthritis. Clin. Orthop. Relat. Res. 474, 342. Atkins, P.R., Fiorentino, N.M., Aoki, S.K., Peters, C.L., Maak, T.G., Anderson, A.E., 2017. In vivo measurements of the ischiofemoral space in recreationally active participants during dynamic activities: a high-speed dual fluoroscopy study. Am. J. Sports Med. 45, 2901–2910. Atkins, P.R., Kobayashi, E.F., Anderson, A.E., Aoki, S.K., 2018. Modified false-profile radiograph of the hip provides better visualization of the anterosuperior femoral head-neck junction. Arthrosc. J. Arthrosc. Relat. Surg. 34, 1236–1243. Audenaert, E.A., Mahieu, P., Pattyn, C., 2011. Three-dimensional assessment of cam engagement in femoroacetabular impingement. Arthrosc. J. Arthrosc. Relat. Surg. 27, 167– 171. Banerjee, P., Mclean, C.R., 2011. Femoroacetabular impingement: a review of diagnosis and management. Curr. Rev. Musculoskelet. Med.4, 23-32. Barros, H.J.M., Camanho, G.L., Bernabé, A.C., Rodrigues, M.B., Leme, L.E.G., 2010. Femoral head-neck junction deformity is related to osteoarthritis of the hip. Clin. Orthop. Relat. Res. 468, 1920–5. Baumgart, C., Hoppe, M.W., Freiwald, J., 2017. Phase-specific ground reaction force analyses of bilateral and unilateral jumps in patients with ACL reconstruction. Orthop. J. Sport. Med. 5, 232596711771091. 51 Bedi, A., Dolan, M., Hetsroni, I., Magennis, E., Lipman, J., Buly, R., Kelly, B.T., 2011. Surgical treatment of femoroacetabular impingement improves hip kinematics. Am. J. Sports Med. 39, 43–49. Bergmann, G., Deuretzbacher, G., Heller, M., Graichen, F., Rohlmann, A., Strauss, J., Duda, G.N., 2001. Hip contact forces and gait patterns from routine activities, J. Biomech. 34, 859871. Bey, M.J., Kline, S.K., Tashman, S., Zauel, R., 2008. Accuracy of biplane x-ray imaging combined with model-based tracking for measuring in-vivo patellofemoral joint motion. J. Orthop. Surg. Res. 3, 38. Bey, M.J., Zauel, R., Brock, S.K., Tashman, S., 2006. Validation of a new model-based tracking technique for measuring three-dimensional, in vivo glenohumeral joint kinematics. J. Biomech. Eng. 128, 604–9. Bharam, S., 2006. Labral Tears, Extra-articular injuries, and hip arthroscopy in the athlete. Clin. Sports Med. 25, 279–292. Bingham, J.T., Papannagari, R., Van de Velde, S.K., Gross, C., Gill, T.J., Felson, D.T., Rubash, H.E., Li, G., 2008. In vivo cartilage contact deformation in the healthy human tibiofemoral joint. Rheumatology 47, 1622–1627. Bischof, J.E., Spritzer, C.E., Caputo, A.M., Easley, M.E., DeOrio, J.K., Nunley, J.A., DeFrate, L.E., 2010. In vivo cartilage contact strains in patients with lateral ankle instability. J. Biomech. 43, 2561–2566. Caputo, A.M., Lee, J.Y., Spritzer, C.E., Easley, M.E., DeOrio, J.K., Nunley, J.A., DeFrate, L.E., 2009. In vivo kinematics of the tibiotalar joint after lateral ankle instability. Am. J. Sports Med. 37, 2241–2248. Clohisy, J.C., Knaus, E.R., Hunt, D.M., Lesher, J.M., Harris-Hayes, M., Prather, H., 2009. Clinical presentation of patients with symptomatic anterior hip impingement. Clin. Orthop. Relat. Res. 467, 638–44. Cross, M., Smith, E., Hoy, D., Nolte, S., Ackerman, I., Fransen, M., Bridgett, L., Williams, S., Guillemin, F., Hill, C.L., Laslett, L.L., Jones, G., Cicuttini, F., Osborne, R., Vos, T., Buchbinder, R., Woolf, A., March, L., 2014. The global burden of hip and knee osteoarthritis: estimates from the global burden of disease 2010 study. Ann. Rheum. Dis. 73, 1323–30. Dahlberg, L.E., Grahn, D., Dahlberg, J.E., Thorstensson, C.A., 2016. A Web-Based Platform for Patients With Osteoarthritis of the Hip and Knee: A Pilot Study. JMIR Res. Protoc. 5, e115. Economopoulos, K.J., Milewski, M.D., Hanks, J.B., Hart, J.M., Diduch, D.R., 2014. Radiographic evidence of femoroacetabular impingement in athletes with athletic pubalgia. Sport. Heal. A Multidiscip. Approach 6, 171–177. Englander, Z.A., Martin, J.T., Ganapathy, P.K., Garrett, W.E., DeFrate, L.E., 2018. Automatic registration of MRI-based joint models to high-speed biplanar radiographs for precise quantification of in vivo anterior cruciate ligament deformation during gait. J. Biomech. 81, 36–44. Ferguson, S., Bryant, J.., Ganz, R., Ito, K., 2000. The influence of the acetabular labrum on hip joint cartilage consolidation: a poroelastic finite element model. J. Biomech. 33, 953–960. Fiorentino, N.M., Atkins, P.R., Kutschke, M.J., Foreman, K.B., Anderson, A.E., 2016a. In-vivo quantification of dynamic hip joint center errors and soft tissue artifact. Gait Posture 50, 246–251. 52 Fiorentino, N.M., Atkins, P.R., Kutschke, M.J., Goebel, J.M., Foreman, K.B., Anderson, A.E., 2017. Soft tissue artifact causes significant errors in the calculation of joint angles and range of motion at the hip. Gait Posture 55, 184–190. Fiorentino, N.M., Kutschke, M.J., Atkins, P.R., Foreman, K.B., Kapron, A.L., Anderson, A.E., 2016b. Accuracy of functional and predictive methods to calculate the hip joint center in young non-pathologic asymptomatic adults with dual fluoroscopy as a reference standard. Ann. Biomed. Eng. 44, 2168–2180. Fox, A.S., Bonacci, J., McLean, S.G., Saunders, N., 2017. Efficacy of ACL injury risk screening methods in identifying high-risk landing patterns during a sport-specific task. Scand. J. Med. Sci. Sports 27, 525–534. Ganz, R., Gill, T.J., Gautier, E., Ganz, K., Krügel, N., Berlemann, U., 2001. Surgical dislocation of the adult hip a technique with full access to the femoral head and acetabulum without the risk of avascular necrosis. J. Bone Joint Surg. Br. 83, 1119–24. Ganz, R., Leunig, M., Leunig-Ganz, K., Harris, W.H., 2008. The etiology of osteoarthritis of the hip: an integrated mechanical concept. Clin. Orthop. Relat. Res. 466, 264–72. Ganz, R., Parvizi, J., Beck, M., Leunig, M., Nötzli, H., Siebenrock, K.A., 2003. Femoroacetabular impingement: a cause for osteoarthritis of the hip. Clin. Orthop. Relat. Res. 112–20. Griffin, D.R., Dickenson, E.J., O’Donnell, J., Agricola, R., Awan, T., Beck, M., Clohisy, J.C., Dijkstra, H.P., Falvey, E., Gimpel, M., Hinman, R.S., Hölmich, P., Kassarjian, A., Martin, H.D., Martin, R., Mather, R.C., Philippon, M.J., Reiman, M.P., Takla, A., Thorborg, K., Walker, S., Weir, A., Bennell, K.L., 2016. The Warwick agreement on femoroacetabular impingement syndrome (FAI syndrome): an international consensus statement. Br. J. Sports Med. 50, 1169–76. Groh, M.M., Herrera, J., 2009. A comprehensive review of hip labral tears. Curr. Rev. Musculoskelet. Med. 2, 105–17. Harris, M.D., Anderson, A.E., Henak, C.R., Ellis, B.J., Peters, C.L., Weiss, J.A., 2012. Finite element prediction of cartilage contact stresses in normal human hips. J. Orthop. Res. 30, 1133–9. Harris, W., 1986. Aetiology of osteoarthritis of the hip. Clin. Orthop. Relat. Res. 213, 20–33. Henak, C.R., Abraham, C.L., Peters, C.L., Sanders, R.K., Weiss, J.A., Anderson, A.E., 2014. Computed tomography arthrography with traction in the human hip for three-dimensional reconstruction of cartilage and the acetabular labrum. Clin Radiol 69, 381–391. Henak, C.R., Ellis, B.J., Harris, M.D., Anderson, A.E., Peters, C.L., Weiss, J.A., 2011. Role of the acetabular labrum in load support across the hip joint. J. Biomech. 44, 2201–6. Henak, C.R., Kapron, A.L., Anderson, A.E., Ellis, B.J., Maas, S.A., Weiss, J.A., 2014. Specimenspecific predictions of contact stress under physiological loading in the human hip: validation and sensitivity studies. Biomech. Model. Mechanobiol. 13, 387–400. Hosseini, A., Van de Velde, S.K., Kozanek, M., Gill, T.J., Grodzinsky, A.J., Rubash, H.E., Li, G., 2010. In-vivo time-dependent articular cartilage contact behavior of the tibiofemoral joint. Osteoarthr. Cartil. 18, 909–16. Ito, K., Minka-Ii, M.-A., Leunig, M., Werlen, S., Ganz, R., 2001. Femoroacetabular impingement and the cam-effect. J. Bone Jt. Surg. 83, 171–176. Kapron, A.L., Anderson, A.E., Aoki, S.K., Phillips, L.G., Petron, D.J., Toth, R., Peters, C.L., 2011. 53 Radiographic prevalence of femoroacetabular impingement in collegiate football players. J. Bone Jt. Surgery-American Vol. 93, e111(1)-e111(10). Kapron, A.L., Anderson, A.E., Peters, C.L., Phillips, L.G., Stoddard, G.J., Petron, D.J., Toth, R., Aoki, S.K., 2012. Hip internal rotation is correlated to radiographic findings of cam femoroacetabular impingement in collegiate football players. YJARS 28, 1661–1670. Kapron, A.L., Aoki, S.K., Peters, C.L., Anderson, A.E., 2014. Subject-specific patterns of femurlabrum contact are complex and vary in asymptomatic hips and hips with femoroacetabular impingement. Clin. Orthop. Relat. Res. 472, 3912–3922. Kapron, A.L, Aoki, S.K., Peters, C.L., Maas, S.A., Bey, M.J., Zauel, R., Anderson, A.E., 2014. Accuracy and feasibility of dual fluoroscopy and model-based tracking to quantify in vivo hip kinematics during clinical exams. J. Appl. Biomech. 30, 461–70. Kapron, A.L., Peters, C.L., Aoki, S.K., Beckmann, J.T., Erickson, J.A., Anderson, M.B., Pelt, C.E., 2015. The Prevalence of radiographic findings of structural hip deformities in female collegiate athletes. Am. J. Sports Med. 43, 1324–1330. Karns, M.R., Patel, S.H., Kolaczko, J., Liu, R.W., Mather, R.C., White, B.J., Nho, S.J., Salata, M.J., 2016. Acetabular rim length: an anatomical study to determine reasonable graft sizes for labral reconstruction. J. Hip Preserv. Surg. 4, hnw038. Klennert, B.J., Ellis, B.J., Maak, T.G., Kapron, A.L., Weiss, J.A., 2017. The mechanics of focal chondral defects in the hip. J. Biomech. 52, 31–37. Knight, S.J., Abraham, C.L., Peters, C.L., Weiss, J.A., Anderson, A.E., 2017. Changes in chondrolabral mechanics, coverage, and congruency following peri-acetabular osteotomy for treatment of acetabular retroversion: A patient-specific finite element study. J. Orthop. Res. 35, 2567–2576. Knight, S.R., Aujla, R., Biswas, S.P., 2011. Total Hip Arthroplasty - over 100 years of operative history. Orthop. Rev. (Pavia). 3, e16. Kubota, S., Inaba, Y., Kobayashi, N., Choe, H., Tezuka, T., Saito, T., 2017. Comparison of improved range of motion between cam-type femoroacetabular impingement and borderline developmental dysplasia of the hip -evaluation by virtual osteochondroplasty using computer simulation. BMC Musculoskelet. Disord. 18, 417. Larson C.M., Giveans M.R., Samuelson, K.M., 2008. Arthroscopic management of femoroacetabular impingement: early outcomes measures. Arthroscopy. 24, 540-546. Larson C.M., Giveans M.R., Samuelson K.M., Stone R.M., Bedi A., 20014. Arthroscopic hip revision surgery for residual femoroacetabular impingement (FAI): surgical outcomes compared with a matched cohort after primary arthroscopic FAI correction. Am. J. Sports Med. 42, 1785-1790. Larson C.M., Giveans M.R., Stone R.M., 2012. Arthroscopic debridement versus refixation of the acetabular labrum associated with femoroacetabular impingement: mean 3.5-year follow-up. Am. J. Sports Med. 40, 1015-1021. Levy D.M., Kuhns B.D., Chahal J., Philippon M.J., Kelly B.T., Nho S.J., 2016. Hip arthroscopy outcomes with respect to patient acceptable symptomatic state and minimal clinically important difference. Arthroscopy. 32, 1877-1886. Leunig, M., Beck, M., Woo, A., Dora, C., Kerboull, M., Ganz, R., 2003. Acetabular rim degeneration: a constant finding in the aged hip. Clin. Orthop. Relat. Res. 413, 201–7. 54 Lewis, C.L., Sahrmann, S.A., 2006. Acetabular Labral Tears. Phys. Ther. 86, 110–121. Li, G., DeFrate, L.E., Park, S.E., Gill, T.J., Rubash, H.E., 2005. In Vivo Articular Cartilage Contact Kinematics of the Knee. Am. J. Sports Med. 33, 102–107. Li, G., Van de Velde, S.K., Bingham, J.T., 2008. Validation of a non-invasive fluoroscopic imaging technique for the measurement of dynamic knee joint motion. J. Biomech. 41, 1616–1622. Liu, F., Kozanek, M., Hosseini, A., Van de Velde, S.K., Gill, T.J., Rubash, H.E., Li, G., 2010. In vivo tibiofemoral cartilage deformation during the stance phase of gait. J. Biomech. 43, 658– 665. Martin, A.B., Hartman, M., Benson, J., Catlin, A., 2016. National Health Spending In 2014: Faster growth driven by coverage expansion and prescription drug spending. Health Aff. 35, 150– 160. McCarthy, J.C., Noble, P.C., Schuck, M.R., Wright, J., Lee, J., 2001. The role of labral lesions to development of early degenerative hip disease. Clin. Orthop. Relat. Res. 393, 25–37. Murphy, L.B., Helmick, C.G., Schwartz, T.A., Renner, J.B., Tudor, G., Koch, G.G., Dragomir, A.D., Kalsbeek, W.D., Luta, G., Jordan, J.M., 2010. One in four people may develop symptomatic hip osteoarthritis in his or her lifetime. Osteoarthr. Cartil. 18, 1372–9. National Center for Health, 2017. Health, United States, 2016: With chartbook on long-term trends in health. Natl. Cent. Heal. S. 2017. Nötzli, H.P., Wyss, T.F., Stoecklin, C.H., Schmid, M.R., Treiber, K., Hodler, J., 2002. The contour of the femoral head-neck junction as a predictor for the risk of anterior impingement. J. Bone Joint Surg. Br. 84, 556–60. Ollivier, M., Le Corroller, T., Parratte, S., Chabrand, P., Argenson, J.-N., Gagey, O., 2017. Mechanical strains passing through the acetabular labrum modify its shape during hip motion: an anatomical study. Knee Surgery, Sport. Traumatol. Arthrosc. 25, 1967–1974. Pataky, T.C., Vanrenterghem, J., Robinson, M.A., 2015. Zero- vs. one-dimensional, parametric vs. non-parametric, and confidence interval vs. hypothesis testing procedures in onedimensional biomechanical trajectory analysis. J. Biomech. 48, 1277–1285. Petersen, W., Petersen, F., Tillmann, B., 2003. Structure and vascularization of the acetabular labrum with regard to the pathogenesis and healing of labral lesions. Arch. Orthop. Trauma Surg. 123, 283–288. Safran, M.R., Giordano, G., Lindsey, D.P., Gold, G.E., Rosenberg, J., Zaffagnini, S., Giori, N.J., 2011. Strains across the acetabular labrum during hip motion. Am. J. Sports Med. 39, 92– 102. Schuermans, J., Danneels, L., Van Tiggelen, D., Palmans, T., Witvrouw, E., 2017. Proximal neuromuscular control protects against hamstring injuries in male soccer players: a prospective study with electromyography time-series analysis during maximal sprinting. Am. J. Sports Med. 45, 1315–1325. Sloan, M., Premkumar, A., Sheth, N.P., 2018. Projected volume of primary total joint arthroplasty in the U.S., 2014 to 2030. J. Bone Joint Surg. Am. 100, 1455–1460. Sochacki, K.R., Yetter, T.R., Morehouse, H., Delgado, D., Nho, S.J., Harris, J.D., 2018. The risk of impingement with sexual activity in femoroacetabular impingement syndrome due to cam morphology: shape matters. Orthop. J. Sport. Med. 6, 2325967118791790. 55 Todd, J.N., Maak, T.G., Ateshian, G.A., Maas, S.A., Weiss, J.A., 2018. Hip chondrolabral mechanics during activities of daily living: Role of the labrum and interstitial fluid pressurization. J. Biomech. 69, 113–120. Uemura, K., Atkins, P.R., Anderson, A.E., Aoki, S.K., 2019. Do your routine radiographs to diagnose cam femoroacetabular impingement visualize the region of the femoral head-neck junction you intended? Arthroscopy 35, 1796–1806. Van de Velde, S.K., Bingham, J.T., Hosseini, A., Kozanek, M., DeFrate, L.E., Gill, T.J., Li, G., 2009. Increased tibiofemoral cartilage contact deformation in patients with anterior cruciate ligament deficiency. Arthritis Rheum. 60, 3693–3702. Vanrenterghem, J., Venables, E., Pataky, T., Robinson, M.A., 2012. The effect of running speed on knee mechanical loading in females during side cutting. J. Biomech. 45, 2444–2449. Wan, L., de Asla, R.J., Rubash, H.E., Li, G., 2008. In vivo cartilage contact deformation of human ankle joints under full body weight. J. Orthop. Res. 26, 1081–1089. Wan, L., de Asla, R.J., Rubash, H.E., Li, G., 2006. Determination of in-vivo articular cartilage contact areas of human talocrural joint under weightbearing conditions. Osteoarthr. Cartil. 14, 1294–301. Warmenhoven, J., Harrison, A., Robinson, M.A., Vanrenterghem, J., Bargary, N., Smith, R., Cobley, S., Draper, C., Donnelly, C., Pataky, T., 2018. A force profile analysis comparison between functional data analysis, statistical parametric mapping and statistical nonparametric mapping in on-water single sculling. J. Sci. Med. Sport 21, 1100–1105. Weber, M., Renkawitz, T., Voellner, F., Craiovan, B., Greimel, F., Worlicek, M., Grifka, J., Benditz, A., 2018. Revision surgery in total joint replacement is cost-intensive. Biomed Res. Int. 8987104, 1–8. Whyte, E.F., Richter, C., O’connor, S., Moran, K.A., 2018. The effect of high intensity exercise and anticipation on trunk and lower limb biomechanics during a crossover cutting manoeuvre. J. Sports Sci. 36, 889–900. Wu, G., Siegler, S., Allard, P., Kirtley, C., Leardini, A., Rosenbaum, D., Whittle, M., D’Lima, D.D., Cristofolini, L., Witte, H., Schmid, O., Stokes, I., 2002. ISB recommendation on definitions of joint coordinate system of various joints for the reporting of human joint motion—part I: ankle, hip, and spine. J. Biomech. 35, 543–548. Yin, P., Li, J.-S., Kernkamp, W.A., Tsai, T.-Y., Baek, S.-H., Hosseini, A., Lin, L., Tang, P., Li, G., 2017. Analysis of in-vivo articular cartilage contact surface of the knee during a step-up motion. Clin. Biomech. 49, 101–106. Zhang, C., Li, L., Forster, B.B., Kopec, J.A., Ratzlaff, C., Halai, L., Cibere, J., Esdaile, J.M., 2015. Femoroacetabular impingement and osteoarthritis of the hip. Can. Fam. Physician 61, 1055–60.