The development and evaluation of a fully automated markerless motion capture workflow.

This study presented a fully automated deep learning based markerless motion capture workflow and evaluated its performance against marker-based motion capture during overground running, walking and counter movement jumping. Multi-view high speed (200 Hz) image data were collected concurrently with marker-based motion capture (criterion data), permitting a direct comparison between methods. Lower limb kinematic data for 15 participants were computed using 2D pose estimation, our 3D fusion process and OpenSim based inverse kinematics modelling. Results demonstrated high levels of agreement for lower limb joint angles, with mean differences ranging "0.1° - 10.5° for hip (3 DoF) joint rotations, and 0.7° - 3.9° for knee (1 DoF) and ankle (2 DoF) rotations. These differences generally fall within the documented uncertainties of marker-based motion capture, suggesting that our markerless approach could be used for appropriate biomechanics applications. We used an open-source, modular and customisable workflow, allowing for integration with other popular biomechanics tools such as OpenSim. By developing open-source tools, we hope to facilitate the democratisation of markerless motion capture technology and encourage the transparent development of markerless methods. This presents exciting opportunities for biomechanics researchers and practitioners to capture large amounts of high quality, ecologically valid data both in the laboratory and in the wild.

Modifications to the net knee moments lead to the greatest improvements in accelerative sprinting performance: a predictive simulation study.

The current body of sprinting biomechanics literature together with the front-side mechanics coaching framework provide various technique recommendations for improving performance. However, few studies have attempted to systematically explore technique modifications from a performance enhancement perspective. The aims of this investigation were therefore to explore how hypothetical technique modifications affect accelerative sprinting performance and assess whether the hypothetical modifications support the front-side mechanics coaching framework. A three-dimensional musculoskeletal model scaled to an international male sprinter was used in combination with direct collocation optimal control to perform (data-tracking and predictive) simulations of the preliminary steps of accelerative sprinting. The predictive simulations differed in the net joint moments that were left 'free' to change. It was found that the 'knee-free' and 'knee-hip-free' simulations resulted in the greatest performance improvements (13.8% and 21.9%, respectively), due to a greater knee flexor moment around touchdown (e.g., 141.2 vs. 70.5 Nm) and a delayed and greater knee extensor moment during stance (e.g., 188.5 vs. 137.5 Nm). Lastly, the predictive simulations which led to the greatest improvements were also found to not exhibit clear and noticeable front-side mechanics technique, thus the underpinning principles of the coaching framework may not be the only key aspect governing accelerative sprinting.

Performance and Micro-Pacing Strategies in a Freestyle Cross-Country Skiing Distance Race.

This study examined the micro-pacing strategies during a distance freestyle cross-country (XC) skiing competition. Nine female and 10 male highly trained XC skiers wore a GNSS device during a FIS-sanctioned race. The course was ~4900 m; women completed two-laps; men completed three-laps. The course was divided into uphill (S1, S3, S5, S7), downhill (S2, S4, S6, S8), and flat (S9) sections for analyses. Statistical parametric mapping was used to determine the course positions (clusters) where total race time or section time was significantly associated with instantaneous skiing speed. Total race time was associated with instantaneous skiing speed during a cluster in S1 on lap 2 for both sexes (t ≥ 5.899, p ≤ 0.008). The two longest uphill sections (S1; S5) and the flat section (S9) contained clusters where section times were related to instantaneous skiing speed for both sexes (p < 0.05). The fastest woman gained 6.9 s on the slowest woman during a cluster in S1 on lap 1 and 7.3 s during a cluster in S9 on lap 1. The fastest man gained 51.7 s on the slowest man over all clusters in S5 over the 3 laps combined. Compared to skiers with longer total race times, skiers with shorter race times skied with faster instantaneous speeds in some clusters of the uphill sections, as well as on the flat section of the course. This study also identified different relative micro-pacing strategies for women and men during freestyle distance XC skiing races. Finally, statistical parametric mapping analyses can help to identify individual strengths and weaknesses for guiding training programs and optimise competition pacing strategies.

Association between biological maturation and anterior cruciate ligament injury risk factors during cutting.

BACKGROUND: Adolescent females are particularly susceptible to suffering anterior cruciate ligament (ACL) injuries, likely influenced by well-established maturational changes. This study investigated ACL biomechanical injury risk factors and their association with biological maturation in females. METHODS: Thirty-five adolescent females (15±1 year) completed a series of maximum-effort 90° unanticipated cutting maneuvers. Established biomechanical ACL injury risk factors (including external knee abduction moments, knee abduction, hip abduction, knee flexion, ground reaction force) were derived from an optoelectronic motion analysis system and force platforms, with inter-limb asymmetries in these risk factors also computed. Biological maturation (percentage of predicted adult stature) was assessed using validated regression equations, incorporating anthropometric measures of participants and their biological parents. RESULTS: Significant bilateral asymmetries were observed with higher peak external knee abduction moments, higher ground reaction forces and less knee flexion (from 0-18% and 30-39% of contact) during the non-dominant vs. dominant cuts (effect sizes =0.36, 0.63 and 0.50, respectively). Maturation did not appear to influence these asymmetries; however, less hip abduction was observed (e.g., 21-51% of contact for dominant cuts) in more biologically-mature females. CONCLUSIONS: These results highlight a potential maturation-related change in cutting technique that may explain the apparent heightened ACL injury risk in this population. As females mature, training targeted at neuromuscular control of hip abductor (e.g. gluteal) muscle groups could potentially mitigate ACL injury risk.

Can We Modify Maximal Speed Running Posture? Implications for Performance and Hamstring Injury Management.

PURPOSE: Sprint kinematics have been linked to hamstring injury and performance. This study aimed to examine if a specific 6-week multimodal intervention, combining lumbopelvic control and unning technique exercises, induced changes in pelvis and lower-limb kinematics at maximal speed and improved sprint performance. METHODS: Healthy amateur athletes were assigned to a control or intervention group (IG). A sprint test with 3-dimensional kinematic measurements was performed before (PRE) and after (POST) 6 weeks of training. The IG program included 3 weekly sessions integrating coaching, strength and conditioning, and physical therapy approaches (eg, manual therapy, mobility, lumbopelvic control, strength and sprint "front-side mechanics"-oriented drills). RESULTS: Analyses of variance showed no between-group differences at PRE. At POST, intragroup analyses showed PRE-POST differences for the pelvic (sagittal and frontal planes) and thigh kinematics and improved sprint performance (split times) for the IG only. Specifically, IG showed (1) a lower anterior pelvic tilt during the late swing phase, (2) greater pelvic obliquity on the free-leg side during the early swing phase, (3) higher vertical position of the front-leg knee, (4) an increase in thigh angular velocity and thigh retraction velocity, (5) lower between-knees distance at initial contact, and (6) a shorter ground contact duration. The intergroup analysis revealed disparate effects (possibly to very likely) in the most relevant variables investigated. CONCLUSION: The 6-week multimodal training program induced clear pelvic and lower-limb kinematic changes during maximal speed sprinting. These alterations may collectively be associated with reduced risk of muscle strain and were concomitant with significant sprint performance improvement.

Development, evaluation and application of a novel markerless motion analysis system to understand push-start technique in elite skeleton athletes.

This study describes the development, evaluation and application of a computer vision and deep learning system capable of capturing sprinting and skeleton push start step characteristics and mass centre velocities (sled and athlete). Movement data were captured concurrently by a marker-based motion capture system and a custom markerless system. High levels of agreement were found between systems, particularly for spatial based variables (step length error 0.001 ± 0.012 m) while errors for temporal variables (ground contact time and flight time) were on average within ± 1.5 frames of the criterion measures. Comparisons of sprinting and pushing revealed decreased mass centre velocities as a result of pushing the sled but step characteristics were comparable to sprinting when aligned as a function of step velocity. There were large asymmetries between the inside and outside leg during pushing (e.g. 0.22 m mean step length asymmetry) which were not present during sprinting (0.01 m step length asymmetry). The observed asymmetries suggested that force production capabilities during ground contact were compromised for the outside leg. The computer vision based methods tested in this research provide a viable alternative to marker-based motion capture systems. Furthermore, they can be deployed into challenging, real world environments to non-invasively capture data where traditional approaches are infeasible.

The accuracy of several pose estimation methods for 3D joint centre localisation.

Human movement researchers are often restricted to laboratory environments and data capture techniques that are time and/or resource intensive. Markerless pose estimation algorithms show great potential to facilitate large scale movement studies 'in the wild', i.e., outside of the constraints imposed by marker-based motion capture. However, the accuracy of such algorithms has not yet been fully evaluated. We computed 3D joint centre locations using several pre-trained deep-learning based pose estimation methods (OpenPose, AlphaPose, DeepLabCut) and compared to marker-based motion capture. Participants performed walking, running and jumping activities while marker-based motion capture data and multi-camera high speed images (200 Hz) were captured. The pose estimation algorithms were applied to 2D image data and 3D joint centre locations were reconstructed. Pose estimation derived joint centres demonstrated systematic differences at the hip and knee (~ 30-50 mm), most likely due to mislabeling of ground truth data in the training datasets. Where systematic differences were lower, e.g., the ankle, differences of 1-15 mm were observed depending on the activity. Markerless motion capture represents a highly promising emerging technology that could free movement scientists from laboratory environments but 3D joint centre locations are not yet consistently comparable to marker-based motion capture.

Can Markerless Pose Estimation Algorithms Estimate 3D Mass Centre Positions and Velocities during Linear Sprinting Activities?

The ability to accurately and non-invasively measure 3D mass centre positions and their derivatives can provide rich insight into the physical demands of sports training and competition. This study examines a method for non-invasively measuring mass centre velocities using markerless human pose estimation and Kalman smoothing. Marker (Qualysis) and markerless (OpenPose) motion capture data were captured synchronously for sprinting and skeleton push starts. Mass centre positions and velocities derived from raw markerless pose estimation data contained large errors for both sprinting and skeleton pushing (mean ± SD = 0.127 ± 0.943 and -0.197 ± 1.549 m·s-1, respectively). Signal processing methods such as Kalman smoothing substantially reduced the mean error (±SD) in horizontal mass centre velocities (0.041 ± 0.257 m·s-1) during sprinting but the precision remained poor. Applying pose estimation to activities which exhibit unusual body poses (e.g., skeleton pushing) appears to elicit more erroneous results due to poor performance of the pose estimation algorithm. Researchers and practitioners should apply these methods with caution to activities beyond sprinting as pose estimation algorithms may not generalise well to the activity of interest. Retraining the model using activity specific data to produce more specialised networks is therefore recommended.

The effect of biological maturity status on ground reaction force production during sprinting.

Sprint ability develops nonlinearly across childhood and adolescence. However, the underpinning ground reaction force (GRF) production is not fully understood. This study aimed to uncover the kinetic factors that explain these maturation-related sprint performance differences in Japanese boys and girls. A total of 153 untrained schoolchildren (80 boys, 73 girls) performed two 50-m maximal effort sprints over a 52-force-platform system embedded in an indoor track. Maturity offset (years from peak height velocity; PHV) was estimated using anthropometric data and used to categorise the children into six-year-long maturation groups (from group 1 [5.5-4.5 years before PHV] to group 6 [0.5 years before to 0.5 years after PHV). Maximum and mean step-averaged velocities across 26 steps were compared across consecutive maturation groups, with further GRF analysis (means and waveforms [statistical parametric mapping]) performed when velocity differences were observed. For boys, higher maximum velocities (effect size ± 90% CI = 1.63 ± 0.69) were observed in maturation group 2 (4.5-3.5 years before PHV) compared to group 1 (5.5-4.5 years before PHV), primarily attributable to higher antero-posterior GRFs across shorter ground contacts. Maximum velocities increased from maturation group 4 (2.5-1.5 years before PHV) to group 5 (1.5-0.5 years before PHV) in the girls (effect size ± 90% CI = 1.00 ± 0.78), due to longer ground contacts rather than higher GRFs per se. Waveform analyses revealed more effective reversal of braking forces and higher propulsive forces (e.g. 14%-77% of stance 4), particularly for comparisons involving boys, which suggested potentially enhanced stretch-shortening ability. Youth sport practitioners should consider these maturation-specific alterations when evaluating young athletes' sprint abilities.

How sprinters accelerate beyond the velocity plateau of soccer players: Waveform analysis of ground reaction forces.

Forces applied to the ground during sprinting are vital to performance. This study aimed to understand how specific aspects of ground reaction force waveforms allow some individuals to continue to accelerate beyond the velocity plateau of others. Twenty-eight male sprint specialists and 24 male soccer players performed maximal-effort 60-m sprints. A 54-force-plate system captured ground reaction forces, which were used to calculate horizontal velocity profiles. Touchdown velocities of steps were matched (8.00, 8.25, and 8.50 m/s), and the subsequent ground contact forces were analyzed. Mean forces were compared across groups and statistical parametric mapping (t tests) assessed for differences between entire force waveforms. When individuals contacted the ground with matched horizontal velocity, ground contact durations were similar. Despite this, sprinters produced higher average horizontal power (15.7-17.9 W/kg) than the soccer players (7.9-11.9 W/kg). Force waveforms did not differ in the initial braking phase (0%-~20% of stance). However, sprinters attenuated eccentric force more in the late braking phase and produced a higher antero-posterior component of force across the majority of the propulsive phase, for example, from 31%-82% and 92%-100% of stance at 8.5 m/s. At this velocity, resultant forces were also higher (33%-83% and 86%-100% of stance) and the force vector was more horizontally orientated (30%-60% and 95%-98% of stance) in the sprinters. These findings illustrate the mechanisms which allowed the sprinters to continue accelerating beyond the soccer players' velocity plateau. Moreover, these force production demands provide new insight regarding athletes' strength and technique training requirements to improve acceleration at high velocity.

A Review of the Evolution of Vision-Based Motion Analysis and the Integration of Advanced Computer Vision Methods Towards Developing a Markerless System.

BACKGROUND: The study of human movement within sports biomechanics and rehabilitation settings has made considerable progress over recent decades. However, developing a motion analysis system that collects accurate kinematic data in a timely, unobtrusive and externally valid manner remains an open challenge. MAIN BODY: This narrative review considers the evolution of methods for extracting kinematic information from images, observing how technology has progressed from laborious manual approaches to optoelectronic marker-based systems. The motion analysis systems which are currently most widely used in sports biomechanics and rehabilitation do not allow kinematic data to be collected automatically without the attachment of markers, controlled conditions and/or extensive processing times. These limitations can obstruct the routine use of motion capture in normal training or rehabilitation environments, and there is a clear desire for the development of automatic markerless systems. Such technology is emerging, often driven by the needs of the entertainment industry, and utilising many of the latest trends in computer vision and machine learning. However, the accuracy and practicality of these systems has yet to be fully scrutinised, meaning such markerless systems are not currently in widespread use within biomechanics. CONCLUSIONS: This review aims to introduce the key state-of-the-art in markerless motion capture research from computer vision that is likely to have a future impact in biomechanics, while considering the challenges with accuracy and robustness that are yet to be addressed.

Textile Electrodes Embedded in Clothing: A Practical Alternative to Traditional Surface Electromyography when Assessing Muscle Excitation during Functional Movements.

Textile electromyography (EMG) electrodes embedded in clothing allow muscle excitation to be recorded in previously inaccessible settings; however, their ability to accurately and reliably measure EMG during dynamic tasks remains largely unexplored. To quantify the validity and reliability of textile electrodes, 16 recreationally active males completed two identical testing sessions, within which three functional movements (run, cycle and squat) were performed twice: once wearing EMG shorts (measuring quadriceps, hamstrings and gluteals myoelectric activity) and once with surface EMG electrodes attached to the vastus lateralis, biceps femoris and gluteus maximus. EMG signals were identically processed to provide average rectified EMG (normalized to walking) and excitation length. Results were compared across measurement systems and demonstrated good agreement between the magnitude of muscle excitation when EMG activity was lower, but agreement was poorer when excitation was higher. The length of excitation bursts was consistently longer when measured using textile vs. surface EMG electrodes. Comparable between-session (day-to-day) repeatability was found for average rectified EMG (mean coefficient of variation, CV: 42.6 and 41.2%) and excitation length (CV: 12.9 and 9.8%) when using textile and surface EMG, respectively. Additionally, similar within-session repeatability (CV) was recorded for average rectified EMG (13.8 and 14.1%) and excitation length (13.0 and 12.7%) for textile and surface electrodes, respectively. Generally, textile EMG electrodes appear to be capable of providing comparable muscle excitation information and reproducibility to surface EMG during dynamic tasks. Textile EMG shorts could therefore be a practical alternative to traditional laboratory-based methods allowing muscle excitation information to be collected in more externally-valid training environments.

The effect of altering loading distance on skeleton start performance: Is higher pre-load velocity always beneficial?

Athletes initiating skeleton runs differ in the number of steps taken before loading the sled. We aimed to understand how experimentally modifying loading distance influenced sled velocity and overall start performance. Ten athletes (five elite, five talent; 67% of all national athletes) underwent two to four sessions, consisting of two dry-land push-starts in each of three conditions (preferred, long and short loading distances). A magnet encoder on the sled wheel provided velocity profiles and the overall performance measure (sled acceleration index). Longer pre-load distances (12% average increase from preferred to long distances) were related to higher pre-load velocity (r = 0.94), but lower load effectiveness (r = -0.75; average reduction 29%). Performance evaluations across conditions revealed that elite athletes' preferred distance push-starts were typically superior to the other conditions. Short loading distances were generally detrimental, whereas pushing the sled further improved some talent-squad athletes' performance. Thus, an important trade-off between generating high pre-load velocity and loading effectively was revealed, which coaches should consider when encouraging athletes to load later. This novel intervention study conducted within a real-world training setting has demonstrated the scope to enhance push-start performance by altering loading distance, particularly in developing athletes with less extensive training experience.

Training-Related Changes in Force-Power Profiles: Implications for the Skeleton Start.

PURPOSE: Athletes' force-power characteristics influence sled velocity during the skeleton start, which is a crucial determinant of performance. This study characterized force-power profile changes across an 18-month period and investigated the associations between these changes and start performance. METHODS: Seven elite- and 5 talent-squad skeleton athletes' (representing 80% of registered athletes in the country) force-power profiles and dry-land push-track performances were assessed at multiple time points over two 6-month training periods and one 5-month competition season. Force-power profiles were evaluated using an incremental leg-press test (Keiser A420), and 15-m sled velocity was recorded using photocells. RESULTS: Across the initial maximum strength development phases, increases in maximum force (Fmax) and decreases in maximum velocity (Vmax) were typically observed. These changes were greater for talent (23.6% and -12.5%, respectively) compared with elite (6.1% and -7.6%, respectively) athletes. Conversely, decreases in Fmax (elite -6.7% and talent -10.3%) and increases in Vmax (elite 8.1% and talent 7.7%) were observed across the winter period, regardless of whether athletes were competing (elite) or accumulating sliding experience (talent). When the training emphasis shifted toward higher-velocity, sprint-based exercises in the second training season, force-power profiles seemed to become more velocity oriented (higher Vmax and more negative force-velocity gradient), which was associated with greater improvements in sled velocity (r = .42 and -.45, respectively). CONCLUSIONS: These unique findings demonstrate the scope to influence force-power-generating capabilities in well-trained skeleton athletes across different training phases. To enhance start performance, it seems important to place particular emphasis on increasing maximum muscle-contraction velocity.

Skeleton sled velocity profiles: a novel approach to understand critical aspects of the elite athletes' start phases.

The development of velocity across the skeleton start is critical to performance, yet poorly understood. We aimed to understand which components of the sled velocity profile determine performance and how physical abilities influence these components. Thirteen well-trained skeleton athletes (>85% of athletes in the country) performed dry-land push-starts alongside countermovement jump and sprint tests at multiple time-points. A magnet encoder attached to the sled wheel provided velocity profiles, which were characterised using novel performance descriptors. Stepwise regression revealed four variables (pre-load velocity, pre-load distance, load effectiveness, velocity drop) to explain 99% variance in performance (ß weights: 1.70, -0.81, 0.25, -0.07, respectively). Sprint times and jump ability were associated (r ± 90% CI) with pre-load velocity (-0.70 ± 0.27 and 0.88 ± 0.14, respectively) and distance (-0.48 ± 0.39 and 0.67 ± 0.29, respectively), however, unclear relationships between both physical measures and load effectiveness (0.33 ± 0.44 and -0.35 ± 0.48, respectively) were observed. Athletes should develop accelerative ability to attain higher velocity earlier on the track. Additionally, the loading phase should not be overlooked and may be more influenced by technique than physical factors. Future studies should utilise this novel approach when evaluating skeleton starts or interventions to enhance performance.

Physical Predictors of Elite Skeleton Start Performance.

PURPOSE: An extensive battery of physical tests is typically employed to evaluate athletic status and/or development, often resulting in a multitude of output variables. The authors aimed to identify independent physical predictors of elite skeleton start performance to overcome the general problem of practitioners employing multiple tests with little knowledge of their predictive utility. METHODS: Multiple 2-d testing sessions were undertaken by 13 high-level skeleton athletes across a 24-wk training season and consisted of flexibility, dry-land push-track, sprint, countermovement-jump, and leg-press tests. To reduce the large number of output variables to independent factors, principal-component analysis (PCA) was conducted. The variable most strongly correlated to each component was entered into a stepwise multiple-regression analysis, and K-fold validation assessed model stability. RESULTS: PCA revealed 3 components underlying the physical variables: sprint ability, lower-limb power, and strength-power characteristics. Three variables that represented these components (unresisted 15-m sprint time, 0-kg jump height, and leg-press force at peak power, respectively) significantly contributed (P < .01) to the prediction (R2 = .86, 1.52% standard error of estimate) of start performance (15-m sled velocity). Finally, the K-fold validation revealed the model to be stable (predicted vs actual R2 = .77; 1.97% standard error of estimate). CONCLUSIONS: Only 3 physical-test scores were needed to obtain a valid and stable prediction of skeleton start ability. This method of isolating independent physical variables underlying performance could improve the validity and efficiency of athlete monitoring, potentially benefitting sport scientists, coaches, and athletes alike.

Detecting meaningful body composition changes in athletes using dual-energy x-ray absorptiometry.

Dual-energy x-ray absorptiometry (DXA) imaging is considered to provide a valid and reliable estimation of body composition when stringent scanning protocols are adopted. However, applied practitioners are not always able to achieve this level of control and the subsequent impact on measurement precision is not always taken into account when evaluating longitudinal body composition changes. The primary aim of this study was to establish the reliability of DXA in an applied elite sport setting to investigate whether real body composition changes can be detected. Additionally, the performance implications of these changes during the training year were investigated. Forty-eight well-trained athletes (from four diverse sports) underwent two DXA scans using a 'real-world' approach (with limited pre-scan controls), typically within 48 h, to quantify typical error of measurement (TEM). Twenty-five athletes underwent further scans, before and after specific training and competition blocks. 'True' body composition changes were evaluated using 2 × TEM thresholds. Twelve bob skeleton athletes also performed countermovement jump and leg press tests at each time point. Many 'true' body composition changes were detected and coincided with the primary training emphases (e.g. lean mass gains during hypertrophy-based training). Clear relationships (r ± 90% CI) were observed between performance changes (countermovement jump and leg press) and changes in lean mass (0.53 ± 0.26 and 0.35 ± 0.28, respectively) and fat mass (-0.44 ± 0.27 and -0.37 ± 0.28, respectively). DXA was able to detect real body composition changes without the use of stringent scanning controls. Associations between changes in body composition and performance demonstrated the potential influence of these changes on strength and power indices.