• upper limb
• prosthetics
• finite-element modeling
• muscle contraction

In this thesis, a finite element model of forearm skeletal muscle has been developed and the obtained mechanical behavior has been investigated. The final goal of the work is to support the implementation of a new human-machine interface for the control of hand prostheses, termed myokinetic control interface. This new approach concerns the implant of multiple permanent magnets into residual muscles following upper limb amputation. Magnet positions are tracked with magnetic sensors located in the socket and the tracking result can be used to control an upper limb prosthesis. In order to test the localization system with realistic magnet trajectories, it is necessary to investigate muscle mechanics, by developing a finite element model of forearm muscles.

Although many researchers have implemented finite element muscle models, a thorough search of the relevant literature did not yield any information about the development of forearm muscle models. This aspect is particularly important for applications related to upper limb prostheses. In fact, the majority of forearm muscles are unipennate, having a fiber architecture that highly influences muscle mechanics and that must be considered in the estimation of magnet displacements.

In order to develop a forearm muscle model, it was first necessary to introduce the anatomical and physiological fundamentals of the musculoskeletal system, with particular attention to the upper limb. A review of constitutive muscle models has been presented, highlighting the main implementations of active muscle contraction. Appropriate constitutive equations for the skeletal muscle have been selected. Those equations, presented by Riccobelli and Ambrosi (2019), allow to model a homogeneous, incompressible and hyperelastic material, which well approximates a skeletal muscles behavior. Two solid phases characterize its structure: an isotropic part, modeling passive material properties, and an active one, describing the fiber contribution. This model has been selected due to its reduced number of parameters and its implementation of the active strain approach. The behavior of such material has thus been implemented into the framework of nonlinear mechanics in COMSOL Multiphysics 5.5.

Geometric and architectural properties of a human forearm muscle have been represented, combining information from ultrasonic images with a 3D CAD model of human forearm, obtained from magnetic resonance images. In particular, a unipennate architecture has been modeled, as it characterizes the majority of forearm muscles. The importance of different boundary conditions, simulating tendons and surrounding connective structures, have been investigated, first on a simplified unipennate model, then on the geometric model of a healthy flexor carpi ulnaris, an important wrist flexor.

Muscle isometric contractions have been simulated, evaluating the model dynamic and kinematic response. These results have then been compared with in vivo data, obtained analyzing muscle response of eight subjects during isometric contraction. The experiments involved two variables for the validation: the force developed and the variation in pennation angle along the muscle. Those quantities have been monitored during maximum voluntary isometric contractions, using a load cell, sEMG electrodes and an ultrasound imaging system. The first two instruments have been used to measure the wrist flexor moment and decouple the contribution of flexor carpi radialis (the other wrist flexor present in the forearm) from the one of flexor carpi ulnaris. Ultrasonography has been used to monitor the variation in pennation angle during those isometric contractions, in three different regions along the muscle: proximally, in the middle and distally.

A parametric study has been performed to evaluate the influence of different parameters on the muscle response. Once identified the correct values able to replicate the muscle behavior of healthy subjects, the same constitutive model has been applied to other two geometries simulating distal and proximal amputations. These configurations have been obtained by cutting the healthy model at the level of the first and second distal third of forearm, respectively. The exploratory response of these models has been investigated to estimate transected muscle behavior, in terms of displacement field.

Finally, regions undergoing maximum and minimum displacement have been identified: their selection as possible implantation sites has been discussed, by studying the trajectories of virtually implanted magnets and comparing results from different simulations.