Background: Stroke contributes significantly to the global rates of disability and mortality. A widely recognised consequence of a stroke is motor impairment, including upper extremity impairments. Rehabilitation has a significant role in the improvement of upper extremity functioning. However, the field of rehabilitation care is currently confronted with several challenges. Telerehabilitation, which includes wearable motion sensors such as Inertial Measurement Units (IMUs), offers a promising solution to improve rehabilitation care in a home-based environment. Therefore, the ArmCoach4Stroke (AC4S) project, initiated by the Erasmus Medical Centre and Rijndam Rehabilitation Institute, was established to develop and evaluate a wearable system for upper extremity rehabilitation at home for post-stroke patients.

Aim: As part of the AC4S project, the aim of this thesis was to investigate the test-retest reliability and criterion validity of an IMU-sensor based method, compared to an optoelectronic system, for measuring kinematic metrics of the upper extremity in post-stroke patients.

Methods: The data from two protocols were subjected to analysis: a reliability and validity protocol. During both protocols, participants were equipped with IMUs and executed predefined calibration movements, as well as reaching and lifting exercises. For the validity protocol, patients also wore markers of the gold standard marker-based Vicon system. The measurements of the validity protocol were performed in a laboratory setting, whereas the reliability protocol simulated measurements in an ecological environment. The indices of calibration movements and exercises were selected manually. The orientations of the IMU sensor were estimated and joint angle metrics were calculated for the reaching and reach-and-return phase of movements. A statistical analysis of the metrics was performed to investigate test-retest reliability and criterion validity separately. This involved the use of the intraclass correlation coefficients (ICC) and Spearman correlation coefficients, respectively.

Results: Eventually, 18 participants were analysed for the reliability protocol and 7 for the validity protocol. The reliability protocol yielded ICCs indicating both good reliability (values >0.75) and poor reliability (values <0.5) for some metrics. Significant test-retest differences were identified for certain metrics. The Spearman correlation coefficients for the validity protocol indicated strong correlations (values >0.7). However, the p-values were above the level of significance. The results further demonstrated considerable differences in absolute values between IMU and Vicon joint angles across various metrics.

Conclusion: A large variability and poor-moderate reliability were observed for multiple metrics, indicating underlying issues with the measurements or metric calculations. Substantial differences in absolute joint angle values were evident, not only between Vicon and IMU, but also between test-retest sessions or between patients. This indicates suboptimal orientation estimation, angle definitions or other fundamental problems that need further investigation.

Accurate kinematic analysis in speed skating is crucial to understand and improve skaters’ unique technique during training. Few research methods have captured joint kinematics in the past, using Inertial Measurement Units (IMUs) or manual annotation of filmed data or marker-based motion capture methods. The large motion capture volume and ice-rink environment hinders these methods to be actively adopted on rink. Thus, with growing accuracy in kinematic estimation algorithms, demand for a biomechanically accurate dataset is high. In this research, we aim to generate a virtual video dataset called ODAH-SpeedSkater, from experimental motion capture data. First, we examine the impact and accuracy of markers during motion capture. After selection of accurate inverse kinematics data, we use the SMPL-X human body model to achieve individual skater body shape and pose throughout the motion, rigging it to a skater-specific scaled OpenSim skeletal model. We render the finalized mesh sequences in an ice rink scene with skater outfits through realistic camera set-ups and configurations. Thus, we successfully create a dataset of 1,326 biomechanically annotated virtual videos of speed skating. Finally, we test our dataset on a pre-trained 3D kinematic estimation algorithm to evaluate its performance on speed skating data. In spite of limited testing, we conclude that training a network exclusively on our dataset may improve its performance, with the ultimate goal of actively implementing such networks in the rink.

In biomechanics, human movement studies are carried out to assess the subject’s kinematic and kinetic variables for a healthy gait. Currently, marker-based systems are the standardized method to extract the kinematic variables of subjects. The marker-based systems pose some serious challenges like cost and portability, and the calibration and synchronization of multiple cameras and sensors are among the other practical challenges. The AI technique often referred as markerless pose estimation methods can overcome these challenges and aid biomechanists and clinicians. Thus, there is a need to develop new deep-learning models that can regress the musculoskeletal model directly from images and videos. However, the deep-learning models are dependent on the quality and quantity of training data. In the current scenario, training data for markerless pose estimation are dependent on the redundant marker-based systems and the challenges persist. To aid this, it is necessary to create a statistical human model or a skinned human animated motion from a biomechanical model to build more training data. From the skinned virtual data consisting of realistic movements, deep-learning models can be trained. Therefore, the aim of the research was to build a pipeline to develop a human-animated model from a musculoskeletal model i.e., the OpenSim model. Two different motions such as walking and running are illustrated as qualitative results. The gait pattern for walking and running motions are realistic from both the frontal and sagittal planes. Furthermore, the deep learning model (D3KE) built by Marian et. al was also evaluated on the animated human motions eg. walking motion from the above pipeline to validate the model. The performance of D3KE is evaluated from different planes of camera views and also a comparison between the upper and lower extremities. The evaluation and comparison are based on two metrics MAEangles (Mean Absolute Error of angles, in radians) and MPBLPE (Mean Per Bony Landmark Position Error, in cm). The MAEangles and MPBLPE are better when observed from the frontal plane rather than from the sagittal plane as the plane of view. Also, the joint angles in the upper extremity show better results compared to the lower extremity. Although, the predictions of the joint angles are way off from the ground truth. This opens the way to perform a feasibility study to optimize joint angles by a pixel loss refinement technique. The findings and remarks on the pixel-loss refinement is tabulated as results.

Patients with neuromuscular diseases that are unable to speak, but whose cognitive ability has been maintained, can be benefited from Brain Computer Interfaces (BCIs). The decoding of inner (covert) speech from EEGs consists of one of the state of the art methods that aim to tackle this issue. High variability between subjects, as well as low signal to noise ratio (SNR) undermine the methods used, and introduce the need for computer assisted solutions. Thus, machine learning models as well as large amounts of recorded data are required to design effective algorithms and produce substantial results. In this study, covert vowel classification was performed in a systematic way, by making use of two openly shared databases from literature; the Coretto database, that contains EEG recordings of native Spanish speakers, and the DAIS dataset, which includes EEG recordings of native Dutch speakers. Six classifiers were initially selected to perform 5-class classification: a Random Forest (RF), a k Nearest Neighbours (kNN), a Gaussian Naive Bayers (GNB), a Deep Convolutional Neural Network (DCNN), a Shallow Convolutional Neural Network (SCNN) and a Long Short Term Memory Recurrent Neural Network (LSTM). The DCNN outperformed the other methods, with average intra-subject accuracies of 35% for Coretto and 39% for DAIS (chance level 20%). Afterwards, an Overt versus Covert trials experiment was implemented, to test the limits of overt speech decoding from EEGs. The overt result was slightly higher than covert, with an intra-subject average value of 37.8% for Coretto and 40.5% for DAIS (chance level 20%). Finally, binary classification was performed to identify those pairs of vowels that can be classified more efficiently. Vowels /a/ and /u/ seemed to perform better in average in both datasets (average of 64.8% for Coretto and 64.4% for DAIS with a chance accuracy of 50%). Future work should focus on identifying the useful parts of the EEG recordings, increasing the SNR and the resolution of the electrodes, and defining the most appropriate dictionaries of words/vowels for a BCI. Also, more studies should follow systematic ways of comparisons between datasets, to obtain less ambiguous insights and lead this field to improvements.

People need to have a comfortable experience in vehicles nowadays. However, they are continuously exposed to vibrations from the vehicle. Madymo active human models (AHM) can be used for comfort analysis and to learn how vibrations influence the human body in several aspects. However, existing AHM are very time-consuming due to their complexity, and the correspondence with human comfort data is only reasonable. In order to more effectively analyze motion comfort, a computationally efficient simplified human model (SHM) is developed and validated. The human body model has 36 degrees of freedom (DoF) considering the following segments: pelvis, two thighs, two lower legs, two feet, lower torso, upper torso and head. The model is validated in fore-aft, lateral and vertical vibrations. The model's postural stabilization parameters are tuned manually, by gradient search and grid search in sequence. Manually tuning gives a group of initial values of parameters for further optimization. According to the results, failure of the gradient search illustrates that this optimization problem is non-smooth. At the same time, grid search gives a relatively better result but also shows that the current cost function does not perfectly represent a good fit and needs improvement. A comparison between AHM and the SHM shows that the SHM has a similar or even better fit for most signals while being a largest factor 116 faster. A comparison of parameters between SHM and multibody human models in the literature shows that the structure of models affects the values of the same parameters. A comparison of different time steps illustrates that shorter time step does not necessarily give higher accuracy for this SHM. Suggestions such as adding muscles and changing body shapes are given for further improvement.

Slipped capital femoral epiphysis (SCFE) is a condition in adolescents affecting the proximal femur. Increased mechanical forces result in fracture of the growth plate. Factors increasing the mechanical forces include for example obesity and endocrine disorders. As a result of the fracture a shift of the bone shaft towards the femur head occurs, causing mobility issues and pain. A treatment option for these patients is an intertrochanteric osteotomy with plate fixation. Restoring range of motion and pain reduction are the main aims of the surgery.

This intertrochanteric osteotomy alters the femur anatomy, resulting in a mismatch between bone and conventional plate fixation. These plates do not fit to the reconstructed bone, resulting in high chances of plate failure. The best option for these patients is to use patient-specific designed bone plates for femur fixation. Optimal mechanical properties of the patient-specific plate are important in limiting plate failure. These properties are dependent on design parameters, which can be adapted dependent on patient criteria and anatomy. Investigations towards patient-specific bone plates and the optimal design parameters for efficient and stable fracture fixation is lacking.

The focus of this study is on a specific SCFE patient case treated with corrective osteotomy using a patient-specific Ti-6Al-4V plate and screws for bone fixation and stability. A valid numerical model (i.e., finite element analysis (FEA)) is created, with physiological loading corresponding to two leg stance and walking, for an initial designed (ID) patient-specific bone plate to analyse plate performance. Screw configuration analysis and topology optimization of the ID plate were performed to design an alternative topology optimized (TO) plate. Biomechanical experiments with digital image correlation (DIC) techniques were performed for mechanical strength analysis and validation of the finite element model (FEM).

Both ID and TO plate were able to withstand two leg stance loading conditions according to both FEA and experimental testing. For walking loading conditions the ID plate showed instability, which could result in higher probability of failure in the plating construct. Quasi-static compression experiments showed failure in the most lateral proximal screw in all tests. The failure in the screws was caused by high bending moments. Optimization of the ID plate resulted in an alternative plate design in SCFE treatment. The TO plate consisted of six screws, decreasing plate length by 18.1%. The TO plate had a variable thickness, where thickness was increased around the lateral side of the plate compared to the ID plate.

The steps followed in this study to design a patient-specific implant and evaluation plate mechanical behaviour with FEM and additional experimental testing have shown to be adequate in investigating plate performance and can be used in the evaluation of newly designed patient-specific bone plates in SCFE patients.

Duchenne Muscular Dystrophy (DMD) is a neuromuscular disorder characterized by progressive muscle weakness as well as progressive muscle fat replacement. Absence of functional dystrophin in muscles is believed to cause membrane instability, making the sarcolemma susceptible to damage when subjected to mechanical stress. However, the process leading from dystrophin-absence to muscle degeneration, and ultimately fat replacement, is still not well understood. Therefore, several hypotheses currently co-exist. Interestingly, muscles of DMD patients are not equally affected, and a more or less consistent successive pattern of muscle fat replacement exists in the lower extremities. The aim of this study was to investigate a possible mechanical role in explaining differential muscle fat replacement in the legs of DMD patients. Based on a published preliminary study, it was hypothesized that eccentric contraction, during which a muscle generates force while lengthening, is associated to muscle fat replacement during gait.

3D gait analysis data of a small cross-sectional sample of typically developing children was used to predict muscle contributions during gait in OpenSim. In one analysis, negative work, the time-integral of force multiplied by lengthening velocity, was normalized to account for differences in physiological cross-sectional area (PCSA) and fiber length, and was subsequently tested for a relation with muscle fat replacement. Furthermore, since muscle architecture parameters PCSA and fiber length are associated to muscle fat replacement, separate effects of total negative work and muscle volume (PCSA*fiber length) on muscle fat replacement were also analyzed.

Normalized negative work was significantly associated with muscle fat replacement, where upper leg muscles generally showed higher fat fractions than lower leg muscles. Furthermore, both the effect of total negative work as well as muscle volume on muscle fat replacement were significant. The effect of total negative work on muscle fat replacement was stronger if muscle volume was smaller. In contrast, normalized and total positive work (i.e., concentric contraction) was not significantly associated with muscle fat replacement.

Eccentric contraction during gait helps explain the differential progression of muscle fat replacement seen among leg muscles of DMD patients. The results of this study generate more insight regarding a mechanical role in the pathophysiology of DMD, and may potentially help in generating or improving therapies that focus on prolonged muscle function preservation.

To better understand and predict osteoarthritis, researchers are developing so-called coupled modelling workflows. Coupled workflows convert data from gait analysis studies to subject-specific tissue mechanical response estimations through the use of musculoskeletal and finite element models. The tissue mechanical response inside the joint is thought to play an integral role in the onset and progression of osteoarthritis. To study this, the design of coupled workflow was proposed. This design contained subject-specific gait data which was processed by a musculoskeletal model with a single degree of freedom knee joint. Musculoskeletal output was transferred through an adjusted generic finite element model of the knee to calculate maximum principal stress and shear strain values in the tibial cartilage of the knee. Proper marker placement is crucial for making an accurate assessment of the patient’s function in gait analysis studies. It has been claimed that marker misplacement is the main cause of measurement variability in gait analysis studies. It however remains unclear how potential marker misplacement propagates to the coupled modelling workflow results. To investigate this, in addition to the design of a coupled workflow, a sensitivity analysis was performed. With this sensitivity analysis we tried to answer the following question: How does marker placement of knee joint markers in gait analysis influence the tissue mechanical response calculated by a coupled workflow? For the sensitivity analysis, knee joint marker placements were virtually perturbed along anterior-posterior, proximal-distal and medial-lateral direction, to mimic marker misplacement. Corresponding knee biomechanics were estimated from the perturbed input data in the coupled workflow. Peak maximum principal stress values varied by up to 0.60MPa and peak shear strain varied by up to 0.08% as a result of perturbed knee marker placement. For cumulative stress levels, broader relative ranges were found. Moreover, the results showed that marker placement along the anterior-posterior direction had the greatest influence on corresponding tissue mechanical response estimations. In future studies a standard error of measurement margin is proposed in the assessment of coupled modelling worfklow results. In conclusion, the proposed workfow was relatively easy to build and provided similar tissue mechanical response result to those as reported by more complex models which were more computationally intensive. This implies that in the future, coupled modelling processes may very well be incorporated into the clinical decision-making process for musculoskeletal disorders like osteoarthritis.

Total Hip Arthroplasty (THA) is a procedure where a defective hip joint with pain symptoms or functional impairment is replaced by a hip prosthesis. During this procedure, the optimal hip prosthesis should be chosen and placed in the correct orientation to achieve a stable hip joint. Failure to do so can lead to dislocation, mechanical failure or infection of the hip joint. To improve the outcome of the THA, range of motion (ROM) of the hip joint can be evaluated during the surgery after placement of the hip prosthesis. Current methods of measuring hip ROM have some limitations. For example, there is a low intra- and inter-test reliability between identification of the bony landmarks and goniometric alignment. The goal of this project was to develop an instrument that can measure the ROM during THA without the limitations of current instruments. An instrument was developed that determines its orientation via an inertial measurement unit (IMU). The IMU is connected on the upper leg via a strap and aligned with the bony landmarks via laser beams. A test has been performed to validate the usage of the instrument. The ROM of the hip for flexion and abduction motion was measured with the instrument and compared to a reference measurement made with two video cameras. A significant difference was found: the instrument did not stay aligned with the hip joint. For further development of the instrument, the drift should be minimised and multiple straps should be tested to allow a better fixation. Additionally, the added-value of the laser requires further research.

The track bound sliding sport of Skeleton was permanently added to Winter Olympics programme in 2002. This has led to increased interest in the sport. Engineering has already proved to be a vital contributor to improved performance in the related sport of Bobsleighing. We hope that engineering can do the same for Skeleton. This report describes an attempt at developing a platform to be used as a real-time training simulator for the sport of Skeleton. For a multitude of reasons athletes are, on average limited to a total of two hours of practice and competitive on-track time in any given year. When compared with time spent practising and in competition in most sports, this is extremely low. It is hypothesized that a simulator can augment track time by providing a realistic environment to practise in, even when access to a track is not available. This work is guided by simulators that have been developed for Bobsleighing. The main components are the models to describe the dynamics of the sport, an input method and visualization of the simulation. The main considerations for the dynamic model are of the track surface, the sled and contact between sled and track surface. These models lead to a system of equations which when solved provide accelerations and contact forces. The accelerations are integrated over a fixed time interval to determine changes in velocities, position and orientation. The position and orientation obtained after the integration is passed on to a game engine which provides the user with real-time visual output of the position and orientation along a digitally recreated track surface. A video game controller was chosen to serve as the input device. It has two joysticks, which can be mapped so as to mimic the forces applied by an athlete. A number of descents were performed using this platform both at real-time speed and at a slower speed to give the user, unfamiliar with the sport, a better chance to steer the sled. We were able to consistently reach the exit of curve 2 in real-time speed and curve 4 at the slower play speed before failure of the simulation. In most cases the algorithm used here proves to take lesser time for computation than the chosen integration time step, which is a great sign for future development as we did not make any attempts to optimise its omputation
time. We made an attempt at validation using time elapsed to traverse a certain distance and the sum of magnitude of Lagrangian multipliers. We had poor results with the time elapsed comparison, with simulated runs being 15% slower than competitive descents. While the sum of Lagrangian multipliers showed good relation to expected behaviour. This first attempt was reasonably successful, and we believe that the lessons learnt from this work has brought us one step closer to realizing a training simulator that can be useful to Skeleton athletes.