Contents lists available at ScienceDirect Clinical Biomechanics journal homepage: www.elsevier.com/locate/clinbiomech A discrete element model to predict anatomy of the psoas muscle and path of the tendon: Design implications for total hip arthroplasty E.A. Audenaert a,b,c,d, ⁎ , V. Khanduja b , C. Bauwens d , T. Van Hoof d , C. Pattyn d , G. Steenackers c a Department of Orthopedic Surgery and Traumatology, Ghent University Hospital, Corneel Heymanslaan 10, 9000 Ghent, Belgium b Department of Trauma and Orthopedics, Addenbrooke's Hospital, Cambridge University Hospitals NHS Foundation Trust, Hills Road, Cambridge CB2 0QQ, UK c Department of Electromechanics, Op3Mech research group, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium d Department of Human Structure and Repair, Ghent University, Corneel Heymanslaan 10, 9000 Ghent, Belgium ABSTRACT Background: The accurate estimation of a muscle's line of action is a fundamental requirement in computational modelling. We present a novel anatomical muscle wrapping technique and demonstrate its clinical use on the evaluation of the Psoas muscle mechanics in hip arthroplasty. Methods: A volume preserving, spring model to parameterize muscle anatomy changes during motion is presented. Validation was performed by a CT scan of a cadaver model in multiple positions. The predicted psoas musculotendinous path was compared with the actual imaging findings. In a second stage, psoas kinetics were compared between a conventional versus a resurfacing hip arthroplasty during gait. Findings: Anatomy prediction error was found to be 2.12 mm on average (SD 1.34 mm). When applied to psoas mechanics during walking, the muscle was found to wrap predominantly around the femoral head providing a biomechanically efficient and nearly constant moment arm for flexion during the entire gait cycle. However, this advantage was found to be lost in small diameter hip arthroplasty designs resulting in an important mechanical disadvantage. The moment arm for flexion, was on average 36% (SD 0.03%) lower in the small diameter conventional hip arthroplasty as compared to the large diameter head of the hip resurfacing and this difference was highly significant. (p < 0.001). Interpretation: Despite the shortcomings of an “in silico” and cadaveric study, our findings are in accordance with previous clinical and gait studies. Furthermore, the findings are strongly in favour of large diameter implant designs, warranting their further development and optimisation. 1. Introduction The main movements at the hip during both walking and running consist of a rather stereotyped motion alternating between flexion and extension, extension occurring primarily during the stance phase and flexion mainly during the swing phase of the ipsilateral leg (Andersson et al., 1995; Andersson et al., 1997). As such, the anterior hip flexor muscles and in particular the iliopsoas muscle importantly define hip joint function and stability during gait, running and sitting (Lewis et al., 2009; Philippon, 2001). However, despite the overwhelming evidence on the importance of the iliopsoas muscle during gait and its docu- mented impact on the rehabilitation process following total hip ar- throplasty, detailed mechanical and anatomical studies on the path of the muscle and tendon, and the related moment arms are nearly non- existent (Wakabayashi et al., 2016). This deficit in the literature is in part to be attributed to the technical difficulties in accurate prediction of position and path of muscles that wrap over several anatomical constraints. The accurate estimation of a muscle's line of action is, however, a fundamental requirement in computational modelling of the musculoskeletal system and for the understanding of muscle function during motion. The earliest techniques to represent a muscle path adopted “via” points connecting straight segments (Delp and Loan, 1995; Kruidhof and Pandy, 2006). Further improvements included the definition of passive geometric constraints -e.g. spheres cylinders or ellipsoids- over which line segments were wrapped (Arnold et al., 2000; Audenaert and Audenaert, 2008; Charlton and Johnson, 2001; Garner and Pandy, 2000). These so -called “obstacle-set” methods have been successfully applied in upper and lower limb models, with estimates for muscle lengths and moments comparable to experimental measurements (Vasavada et al., 2008). Probably the most important advantage of this approach is speed of calculation, allowing for real time visualisation of muscle paths during motion. Further, these estimates can be applied as “seeding” position for approximate volumetric wrapping techniques (Kohout et al., 2013) and full 3D finite element models of muscle (Blemker and Delp, 2005; Reynolds et al., 2004). Recently, research has started to focus on more anatomical defini- tions of the muscle paths, using the actual 3D geometry of underlying bony and nearby soft tissue structures as limiting constraints (Desailly et al., 2010; Gao et al., 2002; Hammer et al., 2019; Kohout et al., 2013; https://doi.org/10.1016/j.clinbiomech.2019.09.004 Received 3 June 2019; Accepted 8 September 2019 ⁎ Corresponding author at: Department of Orthopedic Surgery and Traumatology, Ghent University Hospital, Corneel Heymanslaan 10, 9000 Ghent, Belgium E-mail address: emmanuel.audenaert@ugent.be (E.A. Audenaert). Clinical Biomechanics 70 (2019) 186–191 0268-0033/ © 2019 Elsevier Ltd. All rights reserved. T