Background: Inertial measurement units (IMUs) offer the possibility to capture the lower body motions of players of outdoor team sports. However, various sources of error are present when using IMUs: the definition of the body frames, the soft tissue artefact (STA) and the orientation filter. Methods to minimize these errors are currently being used without knowing their exact influence on the various sources of errors. The goal of this study was to present a method to quantify each of the sources of error of an IMU separately. Methods: An optoelectronic system was used as a gold standard. Rigid marker clusters (RMCs) were designed to construct a rigid connection between the IMU and four markers. This allowed for the separate quantification of each of the sources of error. Ten subjects performed nine different football-specific movements, varying both in the type of movement, and in movement intensity. Results: The error of the definition of the body frames (11.3–18.7 deg RMSD), the STA (3.8–9.1 deg RMSD) and the error of the orientation filter (3.0–12.7 deg RMSD) were all quantified separately for each body segment. Conclusions: The error sources of IMU-based motion analysis were quantified separately. This allows future studies to quantify and optimize the effects of error reduction techniques.

PURPOSE: Neuromuscular fatigue is considered to be important in the etiology of hamstring strain injuries in football. Fatigue is assumed to lead to decreases in hamstring contractile strength and changes in sprinting kinematics, which would increase hamstring strain injury risk. Therefore, the aim was to examine the effects of football-specific fatigue on hamstring maximal voluntary torque (MVT) and rate of torque development (RTD), in relation to alterations in sprinting kinematics. METHODS: Ten amateur football players executed a 90-min running-based football match simulation. Before and after every 15 min of simulated play, MVT and RTD of the hamstrings were obtained in addition to the performance and lower body kinematics during a 20-m maximal sprint. Linear mixed models and repeated measurement correlations were used to assess changes over time and common within participant associations between hamstring contractile properties and peak knee extension during the final part of the swing phase, peak hip flexion, peak combined knee extension and hip flexion, and peak joint angular velocities, respectively. RESULTS: Hamstring MVT and sprint performance were significantly reduced by 7.5% and 14.3% at the end of the football match simulation. Unexpectedly, there were no indications for reductions in RTD when MVT decrease was considered. Decreases in hamstring MVT were significantly correlated to decreases in peak knee angle (R = 0.342) and to increases in the peak combined angle (R = -0.251). CONCLUSIONS: During a football match simulation, maximal voluntary isometric hamstring torque declines. This decline is related to greater peak knee extension and peak combined angle during sprint running, which indicates a reduced capacity of the hamstrings to decelerate the lower leg during sprint running with fatigue.

Current athlete monitoring practice in team sports is mainly based on positional data measured by global positioning or local positioning systems. The disadvantage of these measurement systems is that they do not register lower extremity kinematics, which could be a useful measure for identifying injury-risk factors. Rapid development in sensor technology may overcome the limitations of the current measurement systems. With inertial measurement units (IMUs) securely fixed to body segments, sensor fusion algorithms and a biomechanical model, joint kinematics could be estimated. The main purpose of this article is to demonstrate a sensor setup for estimating hip and knee joint kinematics of team sport athletes in the field. Five male subjects (age 22.5 ± 2.1 years; body mass 77.0 ± 3.8 kg; height 184.3 ± 5.2 cm; training experience 15.3 ± 4.8 years) performed a maximal 30-meter linear sprint. Hip and knee joint angles and angular velocities were obtained by five IMUs placed on the pelvis, both thighs and both shanks. Hip angles ranged from 195° (± 8°) extension to 100.5° (± 8°) flexion and knee angles ranged from 168.6° (± 12°) minimal flexion and 62.8° (± 12°) maximal flexion. Furthermore, hip angular velocity ranged between 802.6 °·s^-1 (± 192 °·s^-1) and-674.9 °·s^-1 (± 130 °·s^-1). Knee angular velocity ranged between 1155.9 °·s^-1 (± 200 °·s^-1) and-1208.2 °·s^-1 (± 264 °·s^-1). The sensor setup has been validated and could provide additional information with regard to athlete monitoring in the field. This may help professionals in a daily sports setting to evaluate their training programs, aiming to reduce injury and optimize performance.