Markerless System Research Articles

Validity of artificial intelligence-based markerless motion capture system for clinical gait analysis: Spatiotemporal results in healthy adults and adults with Parkinson’s disease

Markerless motion capture methods are continuously in development to target limitations encountered in marker-, sensor-, or depth-based systems. Previous evaluation of the KinaTrax markerless system was limited by differences in model definitions, gait event methods, and a homogenous subject sample. The purpose of this work was to evaluate the accuracy of spatiotemporal parameters in the markerless system with an updated markerless model, coordinate- and velocity-based gait events, and subjects representing young adult, older adult, and Parkinson’s disease groups. Fifty-seven subjects and 216 trials were included in this analysis. Interclass correlation coefficients showed excellent agreement between the markerless system and a marker-based reference system for all spatial parameters. Temporal variables were similar, except swing time which showed good agreement. Concordance correlation coefficients were similar with all but swing time showing moderate to almost perfect concordance. Bland-Altman bias and limits of agreement (LOA) were small and improved from previous evaluations. Parameters showed similar agreement across coordinate- and velocity-based gait methods with the latter showing generally smaller LOAs. Improvements in spatiotemporal parameters in the present evaluation was due to inclusion of keypoints at the calcanei in the markerless model. Consistency in the calcanei keypoints relative to heel marker placements may improve results further. Similar to previous work, LOAs are within boundaries to detect differences in clinical groups. Results support the use of the markerless system for estimation of spatiotemporal parameters across age and clinical groups, but caution should be taken when generalizing findings due to remaining error in kinematic gait event methods.

Internet-of-Things-Enabled Markerless Running Gait Assessment from a Single Smartphone Camera.

Running gait assessment is essential for the development of technical optimization strategies as well as to inform injury prevention and rehabilitation. Currently, running gait assessment relies on (i) visual assessment, exhibiting subjectivity and limited reliability, or (ii) use of instrumented approaches, which often carry high costs and can be intrusive due to the attachment of equipment to the body. Here, the use of an IoT-enabled markerless computer vision smartphone application based upon Google’s pose estimation model BlazePose was evaluated for running gait assessment for use in low-resource settings. That human pose estimation architecture was used to extract contact time, swing time, step time, knee flexion angle, and foot strike location from a large cohort of runners. The gold-standard Vicon 3D motion capture system was used as a reference. The proposed approach performs robustly, demonstrating good (ICC(2,1) > 0.75) to excellent (ICC(2,1) > 0.90) agreement in all running gait outcomes. Additionally, temporal outcomes exhibit low mean error (0.01−0.014 s) in left foot outcomes. However, there are some discrepancies in right foot outcomes, due to occlusion. This study demonstrates that the proposed low-cost and markerless system provides accurate running gait assessment outcomes. The approach may help routine running gait assessment in low-resource environments.

Comparison of Lower Extremity Joint Moment and Power Estimated by Markerless and Marker-Based Systems during Treadmill Running.

Background: Markerless (ML) motion capture systems have recently become available for biomechanics applications. Evidence has indicated the potential feasibility of using an ML system to analyze lower extremity kinematics. However, no research has examined ML systems’ estimation of the lower extremity joint moments and powers. This study aimed to compare lower extremity joint moments and powers estimated by marker-based (MB) and ML motion capture systems. Methods: Sixteen volunteers ran on a treadmill for 120 s at 3.58 m/s. The kinematic data were simultaneously recorded by 8 infrared cameras and 8 high-resolution video cameras. The force data were recorded via an instrumented treadmill. Results: Greater peak magnitudes for hip extension and flexion moments, knee flexion moment, and ankle plantarflexion moment, along with their joint powers, were observed in the ML system compared to an MB system (p < 0.0001). For example, greater hip extension (MB: 1.42 ± 0.29 vs. ML: 2.27 ± 0.45) and knee flexion (MB: −0.74 vs. ML: −1.17 nm/kg) moments were observed in the late swing phase. Additionally, the ML system’s estimations resulted in significantly smaller peak magnitudes for knee extension moment, along with the knee production power (p < 0.0001). Conclusions: These observations indicate that inconsistent estimates of joint center position and segment center of mass between the two systems may cause differences in the lower extremity joint moments and powers. However, with the progression of pose estimation in the markerless system, future applications can be promising.

Concurrent validity of artificial intelligence-based markerless motion capture for over-ground gait analysis: A study of spatiotemporal parameters

Gait analysis is used in research and clinical environments; yet several limitations exist in current methodologies. Markerless systems, utilizing high-speed video and artificial intelligence, eliminate most limitations encountered in marker-, depth-, or inertial sensor-based systems; however, further development is needed to improve their utility and accessibility in practice. Spatiotemporal parameters from 22 young adults were estimated during over-ground gait. Nine parameters were calculated using events determined from force plate information combined with foot segment tracking and from motion of the foot relative to the sacrum using marker-based and markerless tracking. Two-way mixed effects, single measurement, absolute agreement and relative consistency interclass correlation coefficients, Bland-Altman bias and limits of agreement, and Lin’s concordance correlations were used to examine the validity of parameters from markerless tracking compared to parameters calculated from gait event methods using force plates and marker-based tracking. Gait speed, stride length, step length, cycle time, and step time from the markerless system all showed strong agreement with the force plate method. Other markerless-determined parameters were not as accurate. Differences in stride width are attributable to inconsistencies in foot segment definitions between models; while differences in stance time, swing time, and double limb support time were influenced by gait event methods. Mean differences in gait parameters were smaller than meaningful clinical differences in Parkinson’s disease patients and within ranges of reference values for elderly subjects. Further studies are needed to determine the validity across other patient groups, but results support the continued development of markerless systems for over-ground gait analysis.

Accuracy of a 3D temporal scanning system for gait analysis: Comparative with a marker-based photogrammetry system

BackgroundCombining the accuracy of marker-based stereophotogrammetry and the usability and comfort of markerless human movement analysis is a difficult challenge. 3D temporal scanners are a promising solution, since they provide moving meshes with thousands of vertices that can be used to analyze human movements. Research questionCan a 3D temporal scanner be used as a markerless system for gait analysis with the same accuracy as traditional, marker-based stereophotogrammetry systems? MethodsA comparative study was carried out using a 3D temporal scanner synchronized with a marker-based stereophotogrammetry system. Two gait cycles of twelve healthy adults were measured simultaneously, extracting the positions of key anatomical points from both systems, and using them to analyze the 3D kinematics of the pelvis, right hip and knee joints. Measurement differences of marker positions and joint angles were described by their root mean square. A t-test was performed to rule out instrumental errors, and an F-test to evaluate the amplifications of marker position errors in dynamic conditions. ResultsThe differences in 3D landmark positions were between 1.9 and 2.4 mm in the reference pose. Marker position errors were significantly increased during motion in the medial-lateral and vertical directions. The angle relative errors were between 3% and 43% of the range of motion, with the greatest difference being observed in hip axial rotation. SignificanceThe differences in the results obtained between the 3D temporal scanner and the marker-based system were smaller than the usual errors due to lack of accuracy in the manual positioning of markers on anatomical landmarks and to soft-tissue artefacts. That level of accuracy is greater than other markerless systems, and proves that such technology is a good alternative to traditional, marker-based motion capture.

Clothing condition does not affect meaningful clinical interpretation in markerless motion capture

Markerless motion capture allows whole-body movements to be captured without the need for physical markers to be placed on the body. This enables motion capture analyses to be conducted in more ecologically valid environments. However, the influences of varied clothing on video-based markerless motion capture assessments remain largely unexplored. This study investigated two types of clothing conditions, “Sport” (gym shirt and shorts) and “Street” (unrestricted casual clothing), on gait parameters during overground walking by 29 participants at self-selected speeds using markerless motion capture. Segment lengths, gait spatiotemporal parameters, and lower-limb kinematics were compared between the two clothing conditions. Mean differences in segment length for the forearm, upper arm, thigh, and shank between clothing conditions ranged from 0.2 cm for the forearm to 0.9 cm for the thigh (p < 0.05 for thigh and shank) but below typical marker placement errors (1 – 2 cm). Seven out of 9 gait spatiotemporal parameters demonstrated statistically significant differences between clothing conditions (p < 0.05), however, these differences were approximately ten times smaller than minimal detectable changes in movement-related pathologies including multiple sclerosis and cerebral palsy. Hip, knee, and ankle joint angle root-mean-square deviation values averaged 2.6° and were comparable to previously reported average inter-session variability for this markerless system (2.8°). The results indicate that clothing, a potential limiting factor in markerless motion capture performance, would negligibly alter meaningful clinical interpretations under the conditions investigated.

Running Footwear and Impact Peak Differences in Recreational Runners.

Simple SummaryRunning is a physical activity practiced by many people to maintain good levels of movement. Recreational runners commonly strike the ground with the postero–lateral zone of the foot, which may be associated with a higher biomechanical load on the lower limb, called impact peak. Different running shoes with specific cushioning are available to overcome the biomechanical load, e.g., shoes with a thickness difference between the forefoot and heel parts of the sole, called heel-to-toe drop. Analyzing the running pattern of recreational runners may be challenging because biomechanics laboratories mainly analyze these characteristics in individuals with visible alterations. To overcome these limitations, we employed a 3D markerless system; furthermore, we investigated footwear use. These parameters were studied to understand the behavior of those runners with and without a higher impact peak. Thirty participants underwent a running analysis and a questionnaire about their footwear. The study’s main finding highlighted kinematic and spatiotemporal differences between the runners presenting a higher impact peak and those without it. Furthermore, we observed that runners without an impact peak prefer shoes with a lower heel-to-toe drop, while the other group prefers shoes with a higher heel-to-toe drop. Investigating biomechanics characteristics is essential to reduce possible injury.Running is a physical activity and the investigation of its biomechanical aspects is crucial both to avoid injuries and enhance performance. Recreational runners may be liable to increased stress over the body, particularly to lower limb joints. This study investigates the different running patterns of recreational runners by analyzing characteristics of the footwear impact peak, spatiotemporal, and kinematic parameters among those that present with a peak impact and those that do not, with a 3D markerless system. Thirty recreational runners were divided into two groups: impact peak group (IP) (n = 16) and no impact peak group (n = 14) (n-IP). Kinematic and spatiotemporal parameters showed a large Cohen’s d effect size between the groups. The mean hip flexion was IP 40.40° versus n-IP 32.30° (d = −0.82). Hip extension was IP 30.20° versus n-IP 27.70° (d = −0.58), and ankle dorsiflexion was IP 20.80°, versus n-IP 13.37° (d = −1.17). Stride length was IP 117.90 cm versus n-IP 105.50 cm (d = −0.84). Steps per minute was IP group 170 spm, versus n-IP 163 spm (d = −0.51). The heel-to-toe drop was mainly 10–12 mm for the IP group and 4–6 mm for the n-IP group. Recreational runners whose hip extension is around 40°, ankle dorsiflexion around 20°, and initial foot contact around 14°, may be predisposed to the presence of an impact peak.

Comparison of marker-less and marker-based motion capture for baseball pitching kinematics

ABSTRACT The purpose of this study was to compare baseball pitching kinematics measured with marker-less and marker-based motion capture. Two hundred and seventy-five fastball pitches were captured at 240 Hz simultaneously with a 9-camera marker-less system and a 12-camera marker system. The pitches were thrown by 30 baseball pitchers (age 17.1 ± 3.1 years). Data for each trial were time-synchronised between the two systems using the instant of ball release. Coefficients of Multiple Correlations (CMC) were computed to assess the similarity of waveforms between the two systems. Discrete measurements at foot contact, during arm cocking, and at ball release were compared between the systems using Bland-Altman plots and descriptive statistics. CMC values for the five time series analysed ranged from 0.88 to 0.97, indicating consistency in movement patterns between systems. Biases for discrete measurements ranged in magnitude from 0 to 16 degrees. Standard deviations of the differences between systems ranged from 0 to 14 degrees, while intraclass correlations ranged from 0.64 to 0.92. Thus, the marker-based and marker-less motion capture systems produced similar patterns for baseball pitching kinematics. However, based on the variations between the systems, it is recommended that a database of normative ranges be established for each system.

Fungal insecticidal activity elevated by non-risky markerless overexpression of an endogenous cysteine-free protein gene in Beauveria bassiana.

Fungal insecticides are notorious for slow kill action, an intrinsic trait that can be improved by the genetic engineering of an exogenous or endogenous virulence factor. However, transgenic insecticides expressing exogenous toxin and herbicide-resistant marker genes may cause unexpected ecological risks and are hardly permitted for field release due to strict regulatory hurdles. It is necessary to improve biotechnology that can speed up fungal insect-killing action and exclude ecological risk source. A markerless transformation system of Beauveria bassiana, a main source of wide-spectrum fungal insecticides, was reconstructed based on the fungal uridine auxotrophy (Δura3). The system was applied for overexpression of the small cysteine-free protein (120 amino acids) gene cfp previously characterized as a regulator of the fungal virulence and gene expression. Three cfp-overexpressed strains showed much faster kill action to Galleria mellonella larvae than the parental wild-type via normal cuticle infection but no change in vegetative growth and aerial condition. The faster kill action was achieved due to not only significant increases in conidial adherence to insect cuticle and total activity of secreted cuticle-degrading Pr1 proteases and of antioxidant enzymes crucial for collapse of insect immune defense but acceleration of hemocoel localization, proliferation in vivo and host death from mummification. The markerless system is free of any foreign DNA fragment as a source of ecologic risk and provides a novel biotechnological approach to enhancing fungal insecticidal activity with non-risky endogenous genes and striding over the regulatory hurdles. © 2022 Society of Chemical Industry.

Comparison of Motion Analysis Systems in Tracking Upper Body Movement of Myoelectric Bypass Prosthesis Users.

Current literature lacks a comparative analysis of different motion capture systems for tracking upper limb (UL) movement as individuals perform standard tasks. To better understand the performance of various motion capture systems in quantifying UL movement in the prosthesis user population, this study compares joint angles derived from three systems that vary in cost and motion capture mechanisms: a marker-based system (Vicon), an inertial measurement unit system (Xsens), and a markerless system (Kinect). Ten healthy participants (5F/5M; 29.6 ± 7.1 years) were trained with a TouchBionic i-Limb Ultra myoelectric terminal device mounted on a bypass prosthetic device. Participants were simultaneously recorded with all systems as they performed standardized tasks. Root mean square error and bias values for degrees of freedom in the right elbow, shoulder, neck, and torso were calculated. The IMU system yielded more accurate kinematics for shoulder, neck, and torso angles while the markerless system performed better for the elbow angles. By evaluating the ability of each system to capture kinematic changes of simulated upper limb prosthesis users during a variety of standardized tasks, this study provides insight into the advantages and limitations of using different motion capture technologies for upper limb functional assessment.

Assessment of a novel deep learning-based marker-less motion capture system for gait study

BackgroundMarker-less systems based on digital video cameras and deep learning for gait analysis could have a deep impact in clinical routine. A recently developed system has shown promising results in terms of joint center position but has not been yet evaluated in terms of gait outcomes. Research questionHow does this novel marker-less system compare to a marker-based reference system in terms of clinically relevant gait parameters? MethodsThe deep learning method behind the developed marker-less system was trained on a dedicated dataset consisting of forty-one asymptomatic and pathological subjects each performing ten walking trials. The system could estimate the three-dimensional position of seventeen joint centers or keypoints (e.g., neck, shoulders, hip, knee, and ankles). We evaluated the marker-less system against a marker-based system in terms of differences in joint position (Euclidean distance), detection of gait events (e.g., heel strike and toe-off), spatiotemporal parameters (e.g., step length, time), kinematic parameters (e.g., hip and knee extension-flexion), and inter-trial reliability for kinematic parameters. ResultsThe marker-less system was able to estimate the three-dimensional position of joint centers with a mean difference of 13.1 mm (SD = 10.2 mm). 99% of the estimated gait events were estimated within 10 ms of the corresponding reference values. Estimated spatiotemporal parameters showed zero bias. The mean and standard deviation of the differences of the estimated kinematic parameters varied by parameter (for example, the mean and standard deviation for knee extension flexion angle were −3.0° and 2.7°). Inter-trial reliability of the measured parameters was similar to that of the marker-based references. SignificanceThe developed marker-less system can measure the spatiotemporal parameters within the range of the minimum detectable changes obtained using the marker-based reference system. Moreover, except for hip extension flexion, the system showed promising results in terms of several kinematic parameters.

Markerless vs. Marker-Based Gait Analysis: A Proof of Concept Study.

The analysis of human gait is an important tool in medicine and rehabilitation to evaluate the effects and the progression of neurological diseases resulting in neuromotor disorders. In these fields, the gold standard techniques adopted to perform gait analysis rely on motion capture systems and markers. However, these systems present drawbacks: they are expensive, time consuming and they can affect the naturalness of the motion. For these reasons, in the last few years, considerable effort has been spent to study and implement markerless systems based on videography for gait analysis. Unfortunately, only few studies quantitatively compare the differences between markerless and marker-based systems in 3D settings. This work presented a new RGB video-based markerless system leveraging computer vision and deep learning to perform 3D gait analysis. These results were compared with those obtained by a marker-based motion capture system. To this end, we acquired simultaneously with the two systems a multimodal dataset of 16 people repeatedly walking in an indoor environment. With the two methods we obtained similar spatio-temporal parameters. The joint angles were comparable, except for a slight underestimation of the maximum flexion for ankle and knee angles. Taking together these results highlighted the possibility to adopt markerless technique for gait analysis.

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.

Deep squat analysis: Functional movement screen criteria vs. biomechanical parameters related to mobility (preliminary results)

Kinematic analysis is a useful and widespread tool used in research and clinical biomechanics for the quantification of human movement. Common marker-based optical motion capture systems are time intensive and require highly trained operators to obtain kinematic data. Markerless motion capture systems offer an alternative method for the measurement of kinematic data with several practical benefits. This work compared the kinematics of human gait measured using a deep learning algorithm-based markerless motion capture system to those from a standard marker-based motion capture system. Thirty healthy adult participants walked on a treadmill while data were simultaneously recorded using eight video cameras and seven infrared optical motion capture cameras, providing synchronized markerless and marker-based data for comparison. The average root mean square distance (RMSD) between corresponding joint centers was less than 2.5 cm for all joints except the hip, which was 3.6 cm. Lower limb segment angles relative to the global coordinate system indicated the global segment pose estimates from both systems were very similar, with RMSD of less than 5.5° for all segment angles except those that represent rotations about the long axis of the segment. Lower limb joint angles captured similar patterns for flexion/extension at all joints, ab/adduction at the knee and hip, and toe-in/toe-out at the ankle. These findings indicate that the markerless system would be a suitable alternative technology in cases where the practical benefits of markerless data collection are preferred.

Concurrent assessment of gait kinematics using marker-based and markerless motion capture

Kinematic analysis is a useful and widespread tool used in research and clinical biomechanics for the quantification of human movement. Common marker-based optical motion capture systems are time intensive and require highly trained operators to obtain kinematic data. Markerless motion capture systems offer an alternative method for the measurement of kinematic data with several practical benefits. This work compared the kinematics of human gait measured using a deep learning algorithm-based markerless motion capture system to those from a standard marker-based motion capture system. Thirty healthy adult participants walked on a treadmill while data were simultaneously recorded using eight video cameras and seven infrared optical motion capture cameras, providing synchronized markerless and marker-based data for comparison. The average root mean square distance (RMSD) between corresponding joint centers was less than 2.5 cm for all joints except the hip, which was 3.6 cm. Lower limb segment angles relative to the global coordinate system indicated the global segment pose estimates from both systems were very similar, with RMSD of less than 5.5° for all segment angles except those that represent rotations about the long axis of the segment. Lower limb joint angles captured similar patterns for flexion/extension at all joints, ab/adduction at the knee and hip, and toe-in/toe-out at the ankle. These findings indicate that the markerless system would be a suitable alternative technology in cases where the practical benefits of markerless data collection are preferred.

Evaluating Lower Limb Kinematics Using Microsoft's Kinect: A Simple, Novel Method.

Purpose: Our aim was to evaluate the Microsoft Kinect sensor (MKS) as a markerless system for motion capture and analysis of lower limb motion, compare it with a state-of-the-art marker-based system (MBS), and investigate its accuracy in simultaneously capturing several lower limb joint movements on several planes while participants walked freely. Method: Participants were asked to walk while gait data were simultaneously recorded by both the MKS and the MBS. Software for analysing the Kinect data stream was developed using Microsoft Visual Studio and Kinect for Windows software development kits. Visual three-dimensional (3D) C-Motion software was used to calculate 3D joint angles of the MBS. Deviation of the joint angles calculated by the two systems was calculated using root-mean-square error (RMSE) on the basis of a designated formula. Results: The calculated RMSE average was <5° between the two systems, a level of difference that has practically no clinical significance. Conclusions: Quantitative measurements of the joint angles of the knee and hip can be acquired using one MKS with some accuracy. The system can be advantageous for clinical use, at the pre- and post-treatment stages of rehabilitation, at significantly lower costs. Further evaluation of the MKS should be performed with larger study populations.

Validation of a 3D Markerless System for Gait Analysis Based on OpenPose and Two RGB Webcams

Markerless human motion capture could be a potential tool for studying body biomechanics. Despite the extensive previous studies, the design of an accurate markerless system able to perform a quantitative clinical gait analysis is still an open challenge. Our purpose was to characterize the performance of a low-cost markerless system, consisting of the open-source library OpenPose, two webcams and a linear triangulation algorithm. The system was validated in terms of 3D kinematic gait analysis, by comparison with inertial sensors. Two synchronized videos of six healthy subjects were recorded in three webcam configurations, in walking and running sessions on a treadmill. Sagittal joint angles were compared between the two systems, to evaluate the kinematic performance of the markerless system. Results showed that the angular paths, provided by OpenPose, were close to the ones computed by inertial sensors in all camera positions, but OpenPose inaccuracy affected the computation of joint angles parameters producing absolute errors up to 14.0°± 1.8°. The observed differences depended on OpenPose ability in the recognition of joints centers and on camera configurations. In the optimal condition (joint visible from both cameras during the whole trial) our system accuracy resulted equal to 1.6 °± 0.3 °. This system could be useful in those applications where a real-time body tracking is required, i.e. video games or virtual rehabilitative trials. When a quantitative analysis is needed, the low level of accuracy may be overcome, by both increasing the cameras' number and implementing an optimization triangulation algorithm.

Estimating anatomically plausible segment orientations using a kinect one sensor

Introduction Marker-based motion tracking systems are the golden standard for human motion analysis, however such systems are expensive, non-portable and require long time subject preparation. The Kinect One sensor, being inexpensive, portable and markerless, appears as a reliable and valid alternative to the marker-based systems in several situations [1–3]. This sensor acquires depth image data and colour camera data that are processed by a tracking algorithm to estimate the three-dimensional position of twenty-five anatomical joints in real-time [4]. Nevertheless, the internal orientations of each anatomical segment are poorly estimated. The main objective of this work is to study the effectiveness of vector orthogonalization methods to estimate the relative internal orientations of the anatomical body segments using the skeletal data acquired by a Kinect One sensor. Materials and methods Twenty-eight young healthy adults (25 ± 9 yrs old, 170 ± 9 cm height, 61 ± 9 kg weight, 13 women) performed 5 repetitions of ten different elementary movements: shoulder flexion/hyperextension, shoulder abduction/adduction, shoulder transversal abduction/adduction, shoulder medial/lateral rotation, elbow flexion, forearm pronation/supination, hip flexion/hyperextension, hip abduction/adduction, knee flexion and hip medial/lateral rotation. On each repetition, the subject initiated the movement in an adapted pose of the anatomical reference position and once finished returned to the initial position. Data was collected, simultaneously, using a marker-based system (Qualysis − 100 Hz) and a markerless system (Kinect One − 30 Hz). All participants signed consent forms.The biomechanical model used was composed by eleven anatomical segments: the head, the chest, the abdomen and both arms, forearms, thighs and legs. Six different vector orthogonalization methods (Householder, Eberly, Square Plate, Spherical and Projection Matrix) were used to estimate the relative orientations of the anatomical body segments from Kinect One sensor model [5]. Pearson’s correlation coefficient was used to compare the anatomical body segments orientations of all model segments obtained with both systems. Results The results obtained show that the six techniques implemented present a moderate to high correlation (0.58 − 0.93) between segments longitudinal axis of rotation while for the remaining axes (anterior-posterior and medial-lateral) they show a moderate to negligible correlation (–0.37 to 0.46). Additionally, the performance of each technique varies according the selected movement. For example, the Householder technique presents different correlation values when applied to the following movements, hip flexion (0.84), hip abduction (–0.05), knee flexion (0.78), shoulder flexion (0.36), elbow flexion (0.80) present relevant differences. Discussion and conclusions Although vector orthogonalization techniques are capable to estimate plausible orientations, the results given the same movement shows significant differences, suggesting that not all vector orthogonalization techniques are appropriate for all movements. Therefore, it is necessary to careful select the best technique for each movement in order to obtain valid results. Finally, it is possible to conclude that Kinect One shows good results for some kinematic variables, nevertheless, it needs to improve the precision on the estimation of the joints’ position and all body segments’ orientation in order to obtain results similar to marker-based systems.

A novel dataset and deep learning-based approach for marker-less motion capture during gait

BackgroundThe deep learning-based human pose estimation methods, which can estimate joint centers position, have achieved promising results on the publicly available human pose datasets (e.g., Human3.6 M). However, these datasets may be less efficient for gait study, particularly for clinical applications, because of the limited number of subjects, their homogeneity (all asymptomatic adults), and the errors introduced by marker placement on subjects’ regular clothing. Research questionHow a new human pose dataset, adapted for gait study, could contribute to the advancement and evaluation of marker-less motion capture systems? MethodsA marker-less system, based on deep learning-based pose estimation methods, was proposed. A new dataset (ENSAM dataset) was collected. Twenty-two asymptomatic adults, one adult with scoliosis, one adult with spondylolisthesis, and seven children with bone disease performed ten walking trials, while being recorded both by the proposed marker-less system and a reference system – combining a marker-based motion capture system and a medical imaging system (EOS). The dataset was split into training and test sets. The pose estimation method, already trained on the Human3.6 M dataset, was evaluated on the ENSAM test set, then reevaluated after further training on the ENSAM training set. The joints coordinates were evaluated, using Bland-Altman bias and 95 % confidence interval, and joint position error (the Euclidean distance between the estimated joint centers and the corresponding reference values). ResultsThe Bland-Altman 95 % confidence intervals were substantially improved after finetuning the pose estimation method on the ENSAM training set (e.g., from 106.9 mm to 17.4 mm for the hip joint). With the new dataset and approach, the mean joint position error varied from 6.2 mm for ankles to 21.1 mm for shoulders. SignificanceThe proposed marker-less system achieved promising results in terms of joint position errors. Future studies are necessary to assess the system in terms of gait parameters.

Validation of a Commercially Available Markerless Motion-Capture System for Trunk and Lower Extremity Kinematics During a Jump-Landing Assessment.

Field-based, portable motion-capture systems can be used to help identify individuals at greater risk of lower extremity injury. Microsoft Kinect-based markerless motion-capture systems meet these requirements; however, until recently, these systems were generally not automated, required substantial data postprocessing, and were not commercially available. To validate the kinematic measures of a commercially available markerless motion-capture system. Descriptive laboratory study. Laboratory. A total of 20 healthy, physically active university students (10 males, 10 females; age = 20.50 ± 2.78 years, height = 170.36 ± 9.82 cm, mass = 68.38 ± 10.07 kg, body mass index = 23.50 ± 2.40 kg/m2). Participants completed 5 jump-landing trials. Kinematic data were simultaneously recorded using Kinect-based markerless and stereophotogrammetric motion-capture systems. Sagittal- and frontal-plane trunk, hip-joint, and knee-joint angles were identified at initial ground contact of the jump landing (IC), for the maximum joint angle during the landing phase of the initial landing (MAX), and for the joint-angle displacement from IC to MAX (DSP). Outliers were removed, and data were averaged across trials. We used intraclass correlation coefficients (ICCs [2,1]) to assess intersystem reliability and the paired-samples t test to examine mean differences (α ≤ .05). Agreement existed between the systems (ICC range = -1.52 to 0.96; ICC average = 0.58), with 75.00% (n = 24/32) of the measures being validated (P ≤ .05). Agreement was better for sagittal- (ICC average = 0.84) than frontal- (ICC average = 0.35) plane measures. Agreement was best for MAX (ICC average = 0.77) compared with IC (ICC average = 0.56) and DSP (ICC average = 0.41) measures. Pairwise comparisons identified differences for 18.75% (6/32) of the measures. Fewer differences were observed for sagittal- (0.00%; 0/15) than for frontal- (35.29%; 6/17) plane measures. Between-systems differences were equivalent for MAX (18.18%; 2/11), DSP (18.18%; 2/11), and IC (20.00%; 2/10) measures. The markerless system underestimated sagittal-plane measures (86.67%; 13/15) and overestimated frontal-plane measures (76.47%; 13/17). No trends were observed for overestimating or underestimating IC, MAX, or DSP measures. Moderate agreement existed between markerless and stereophotogrammetric motion-capture systems. Better agreement existed for larger (eg, sagittal-plane, MAX) than for smaller (eg, frontal-plane, IC) joint angles. The DSP angles had the worst agreement. Markerless motion-capture systems may help clinicians identify individuals at greater risk of lower extremity injury.