Effect of various upper limb multibody models on soft tissue artefact correction: A case study

Is an ellipsoid surface suitable to model the scapulothoracic sliding plane?

Closed loop kinematic chain approaches are commonly used to assess scapular kinematics but with heterogeneous ellipsoid calibration procedures. This study aimed to assess whether an ellipsoid surface can model the scapulothoracic sliding plane and determine the optimal number of static poses to calibrate the ellipsoid parameters. An intracortical pin with a rigid cluster of four reflective markers was inserted into the left scapular spine of two healthy males (P1 and P2). They performed arm elevations, internal rotations, ball throwing, hockey shooting, and eating movements. Ellipsoid radii and center location were functionally calibrated for each participant and each movement, either based on all frames of a movement or based on a reduced number of frames (from 3 to 200 equally position-distributed frames). Across both participants and all movements, ellipsoid radii varied up to 10.2 cm, 3.9 cm, and 18.4 cm in the antero-posterior, medio-lateral, and cranio-caudal directions, respectively. When all frames of a movement were considered for calibration, the median scapula-to-ellipsoid distance was, on average, 0.52 mm and 0.38 mm for P1 and P2, respectively. When only five frames were considered for ellipsoid calibration, the scapula-to-ellipsoid median distance slightly increased with 0.57 mm and 0.47 mm for P1 and P2, respectively. To conclude, this study highlights that an ellipsoid surface may effectively be appropriate to model the scapulothoracic sliding plane, especially when the calibration is functional, participant- and movement-specific. Furthermore, the number of poses required for the ellipsoid calibration can be reduced to five, minimizing the experimental cost.

Influence of marker weights optimization on scapular kinematics estimated with a multibody kinematic optimization

Scapular kinematic estimates are altered by soft tissue artefacts, therefore experimental and numerical methods should be developed to improve their accuracy. This study aimed to assess the influence of weights applied to the scapula markers within a closed-loop multibody kinematic optimization on scapular kinematic estimates. Fifteen healthy volunteers performed static postures mimicking analytical, daily living and sport movements. Scapulo-thoracic angles were computed either from a scapula locator as the reference, or from a closed-loop multibody-kinematic optimization (MKO) including a participant-specific point-on-ellipsoid scapulothoracic joint. Weights applied to scapula markers in the MKO were optimized to minimize the difference in scapular orientation from the reference. Optimizing weighting sets significantly (p < 0.0001) improved scapular orientation from 0.9° to 12.1° in comparison to scapular kinematics estimated with non-optimized weighting sets. The mean optimized weighting set contained no neglectable weight for all markers from the acromion to the medial border of the scapular spine but showed no significant difference (p = 0.547) compared to homogeneous weights. Optimized weighting sets were participant- and movement- specific. To conclude, homogenous weights applied on redundant markers located from acromion to scapular medial border spine are recommended when estimating scapular kinematics in upper limb MKO.

Reverse Total Shoulder Arthroplasty Alters Humerothoracic, Scapulothoracic, and Glenohumeral Motion During Weighted Scaption

BACKGROUND

Reverse total shoulder arthroplasty (rTSA) typically restores active arm elevation. Prior studies in patients with rTSA during tasks that load the arm had limitations that obscured underlying three-dimensional (3D) kinematic changes and the origins of motion restrictions. Understanding the scapulothoracic and glenohumeral contributions to loaded arm elevation will uncover where functional deficits arise and inform strategies to improve rTSA outcomes.

QUESTIONS/PURPOSES

In a cohort of patients who had undergone rTSA and a control cohort, we asked: (1) Is there a difference in maximum humerothoracic elevation when scapular plane elevation (scaption) is performed with and without a handheld weight? (2) Is maximum humerothoracic elevation related to factors like demographics, patient-reported outcome scores, isometric strength, and scapular notching (in the rTSA group only)? (3) Are there differences in underlying 3D scapulothoracic and glenohumeral motion during scaption with and without a handheld weight?

METHODS

Ten participants who underwent rTSA (six males, four females; age 73 ± 8 years) were recruited at follow-up visits if they were more than 1 year postoperative (24 ± 11 months), had a BMI less than 35 kg/m 2 (29 ± 4 kg/m 2 ), had a preoperative CT scan, and could perform pain-free scaption. Data from 10 participants with a nonpathologic shoulder, collected previously (five males, five females; age 58 ± 7 years; BMI 26 ± 3 kg/m 2 ), were a control group with the same high-resolution quantitative metrics available for comparison. Participants in both groups performed scaption with and without a 2.2-kg handheld weight while being imaged with biplane fluoroscopy. Maximum humerothoracic elevation and 3D scapulothoracic and glenohumeral kinematics across their achievable ROM were collected via dynamic imaging. In the same session the American Shoulder and Elbow Surgeons (ASES) score, the Simple Shoulder Test (SST), and isometric strength were collected. Data were compared between weighted and unweighted scaption using paired t-tests and linear mixed-effects models.

RESULTS

When compared with unweighted scaption, maximum humerothoracic elevation decreased during weighted scaption for patients who underwent rTSA (-25° ± 30°; p = 0.03) but not for the control group (-2° ± 5°; p = 0.35). In the rTSA group, maximum elevation correlated with the ASES score (r = 0.72; p = 0.02), and weighted scaption correlated with BMI (r = 0.72; p = 0.02) and the SST (r = 0.76; p = 0.01). Scapular notching was observed in three patients after rTSA (Grades 1 and 2). Four of 10 patients who underwent rTSA performed weighted scaption to less than 90° humerothoracic elevation using almost exclusively scapulothoracic motion, with little glenohumeral contribution. This manifested as changes in the estimated coefficient representing mean differences in slopes in the humerothoracic plane of elevation (-12° ± 2°; p < 0.001) and true axial rotation (-16° ± 2°; p < 0.001), scapulothoracic upward rotation (7° ± 1°; p < 0.001), and glenohumeral elevation (-12° ± 1°; p < 0.001), plane of elevation (-8° ± 3°; p = 0.002), and true axial rotation (-11° ± 2°; p < 0.001). The control group demonstrated small differences between scaption activities (< |2°|), but a 10° increase in humerothoracic and glenohumeral axial rotation (both p < 0.001).

CONCLUSION

After rTSA surgery, maximum humerothoracic elevation decreased during weighted scaption by up to 88° compared with unweighted scaption, whereas 4 of 10 patients could not achieve more than 90° of elevation. These patients exhibited appreciable changes in nearly all scapulothoracic and glenohumeral degrees of freedom, most notably a near absence of glenohumeral elevation during weighted scaption. Patients with rTSA have unique strategies to elevate their arms, often with decreased glenohumeral motion and resultant compensation in scapulothoracic motion. In contrast, the control group showed few differences when lifting a handheld weight.

CLINICAL RELEVANCE

Functional deficiency in activities that load the shoulder after rTSA surgery can affect patient independence, and they may be prevalent but not captured in clinical studies. Pre- or postoperative rehabilitation to strengthen scapular stabilizers and the deltoid should be evaluated against postoperative shoulder function. Further study is required to determine the etiology of deficient glenohumeral motion after rTSA, and the most effective surgical and/or rehabilitative strategies to restore deficient glenohumeral motion after rTSA.

Manual wheelchair biomechanics while overcoming various environmental barriers: A systematic review

During manual wheelchair (MWC) locomotion, the user’s upper limbs are subject to heavy stresses and fatigue because the upper body is permanently engaged to propel the MWC. These stresses and fatigue vary according to the environmental barriers encountered outdoors along a given path. This study aimed at conducting a systematic review of the literature assessing the biomechanics of MWC users crossing various situations, which represent physical environmental barriers. Through a systematic search on PubMed, 34 articles were selected and classified according to the investigated environmental barriers: slope; cross-slope; curb; and ground type. For each barrier, biomechanical parameters were divided into four categories: spatiotemporal parameters; kinematics; kinetics; and muscle activity. All results from the different studies were gathered, including numerical data, and assessed with respect to the methodology used in each study. This review sheds light on the fact that certain situations (cross-slopes and curbs) or parameters (kinematics) have scarcely been studied, and that a wider set of situations should be studied. Five recommendations were made at the end of this review process to standardize the procedure when reporting materials, methods, and results for the study of biomechanics of any environmental barrier encountered in MWC locomotion: (i) effectively reporting barriers’ lengths, grades, or heights; (ii) striving for standardization or a report of the approach conditions of the barrier, such as velocity, especially on curbs; (iii) reporting the configuration of the used MWC, and if it was fitted to the subject’s morphology; (iv) reporting rotation sequences for the expression of moments and kinematics, and when used, the definition of the musculoskeletal model; lastly (v) when possible, reporting measurement uncertainties and model reconstruction errors.

Statistical Quantification of the Effects of Marker Misplacement and Soft-Tissue Artifact on Shoulder Kinematics and Kinetics

The assessment of shoulder kinematics and kinetics are commonly undertaken biomechanically and clinically by using rigid-body models and experimental skin-marker trajectories. However, the accuracy of these trajectories is plagued by inherent skin-based marker errors due to marker misplacements (offset) and soft-tissue artifacts (STA). This paper aimed to assess the individual contribution of each of these errors to kinematic and kinetic shoulder outcomes computed using a shoulder rigid-body model. Baseline experimental data of three shoulder planar motions in a young healthy adult were collected. The baseline marker trajectories were then perturbed by simulating typically observed population-based offset and/or STA using a probabilistic Monte-Carlo approach. The perturbed trajectories were then used together with a shoulder rigid-body model to compute shoulder angles and moments and study their accuracy and variability against baseline. Each type of error was studied individually, as well as in combination. On average, shoulder kinematics varied by 3%, 6% and 7% due to offset, STA or combined errors, respectively. Shoulder kinetics varied by 11%, 27% and 28% due to offset, STA or combined errors, respectively. In conclusion, to reduce shoulder kinematic and kinetic errors, one should prioritise reducing STA as they have the largest error contribution compared to marker misplacements.

Subject-specific model-derived kinematics of the shoulder based on skin markers during arm abduction up to 180° - assessment of 4 gleno-humeral joint models

Accuracy of shoulder kinematics predicted by multi-body kinematics optimisation depend on the joint models used. This study assesses the influence of four different subject-specific gleno-humeral joint models within multi-body kinematics optimisation: a 6-degree-of-freedom joint (i.e. single-body kinematics optimisation), a sphere-on-sphere joint (with two spheres of different radii) and a spherical joint with or without penalised translation. To drive these models, the 3D coordinates of 12 skin markers of 6 subjects performing static arm abduction poses up to 180° were used. The reference data was obtained using biplane X-rays from which 3D bone reconstructions were generated: scapula and humerus were 3D reconstructed by fitting a template model made of geometrical primitives on the two bones' X-rays. Without any motion capture system, the recording of the skin markers was performed at the very same time than the X-rays with radiopaque markers. The gleno-humeral displacements and angles, and scapula-thoracic angles were computed. The gleno-humeral sphere-on-sphere joint provided slightly better results than the spherical joint with or without penalised translation, but considerably better gleno-humeral displacements than the 6-DoF joint. Considering that it can easily be personalised from medical images, this sphere-on-sphere model seems promising for shoulder multi-body kinematics optimisation.

Feasibility of an alternative method to estimate glenohumeral joint center from videogrammetry measurements and CT/MRI of patients

Videogrammetry is commonly used to record upper limb motions. However, it cannot track the glenohumeral joint center (GH). GH is required to reconstruct upper limb motions. Therefore, it is often estimated by separately measuring scapular motions using scapular kinematics measurements devices (SKMD). Applications of SKMD are neither straightforward nor always noninvasive. Therefore, this work investigates the feasibility of an alternative method to estimate GH from videogrammetry using a CT/MRI image of subject's glenohumeral joint and without requiring SKMD. In order to evaluate the method's accuracy, its GH estimations were compared to reference GH trajectories. The method was also applied to estimate scapular configurations and reconstruct an abduction motion measured by videogrammetry. The accuracy of GH estimations were within 5 mm, and the reconstructed motion was in good agreement with reported in vivo measurements.

Comparison of shoulder kinematic chain models and their influence on kinematics and kinetics in the study of manual wheelchair propulsion

Several kinematic chains of the upper limbs have been designed in musculoskeletal models to investigate various upper extremity activities, including manual wheelchair propulsion. The aim of our study was to compare the effect of an ellipsoid mobilizer formulation to describe the motion of the scapulothoracic joint with respect to regression-based models on shoulder kinematics, shoulder kinetics and computational time, during manual wheelchair propulsion activities. Ten subjects, familiar with manual wheelchair propulsion, were equipped with reflective markers and performed start-up and propulsion cycles with an instrumented field wheelchair. Kinematic data obtained from the optoelectronic system and kinetic data measured by the sensors on the wheelchair were processed using the OpenSim software with three shoulder joint modeling versions (ellipsoid mobilizer, regression equations or fixed scapula) of an upper-limb musculoskeletal model. As expected, the results obtained with the three versions of the model varied, for both segment kinematics and shoulder kinetics. With respect to the model based on regression equations, the model describing the scapulothoracic joint as an ellipsoid could capture the kinematics of the upper limbs with higher fidelity. In addition, the mobilizer formulation allowed to compute consistent shoulder moments at a low computer processing cost. Further developments should be made to allow a subject-specific definition of the kinematic chain.

On the influence of the shoulder kinematic chain on joint kinematics and musculotendon lengths during wheelchair propulsion estimated from multibody kinematics optimization

Multibody kinematic optimization is frequently used to assess shoulder kinematics during manual wheelchair (MWC) propulsion but multiple kinematics chains are available. It is hypothesized that these different kinematic chains affect marker tracking, shoulder kinematics and resulting musculotendon (MT) lengths. In this study, shoulder kinematics and MT lengths obtained from four shoulder kinematic chains (open-loop thorax-clavicle-scapula-humerus (M1), closed-loop with contact ellipsoid (M2), scapula rhythm from regression equations (M3), and a single ball-and- socket joint between the thorax and the humerus (M4) were compared. Right-side shoulder kinematics from seven subjects were obtained with 34 reflective markers and a scapula locator using an optoelectronic motion capture system while propelling on a MWC simulator. Data was processed based on the four models. Results showed the impact of shoulder kinematic chains on all studied variables. Marker reconstruction errors were found similar between M1 and M2 and lower than for M3 and M4. Few degrees of freedom (DoF) were noticeably different between M1 and M2, but all shoulder DoFs were significantly affected between M1 and M4. As a consequence of differences in joint kinematics, MT lengths were affected by the kinematic chain definition. The contact ellipsoid (M2) was found as a good trade-off between marker tracking and penetration avoidance of the scapula. The regression-based model (M3) was less efficient due to limited humerus elevation during MWC propulsion, as well as the ball-and-socket model (M4) which appeared not suitable for upper limbs activities, including MWC propulsion.

A survey of human shoulder functional kinematic representations

In this survey, we review the field of human shoulder functional kinematic representations. The central question of this review is to evaluate whether the current approaches in shoulder kinematics can meet the high-reliability computational challenge. This challenge is posed by applications such as robot-assisted rehabilitation. Currently, the role of kinematic representations in such applications has been mostly overlooked. Therefore, we have systematically searched and summarised the existing literature on shoulder kinematics. The shoulder is an important functional joint, and its large range of motion (ROM) poses several mathematical and practical challenges. Frequently, in kinematic analysis, the role of the shoulder articulation is approximated to a ball-and-socket joint. Following the high-reliability computational challenge, our review challenges this inappropriate use of reductionism. Therefore, we propose that this challenge could be met by kinematic representations, that are redundant, that use an active interpretation and that emphasise on functional understanding.

Joint Center Estimation Using Single-Frame Optimization: Part 2: Experimentation

Human motion capture is driven by joint center location estimates, and error in their estimation can be compounded by subsequent kinematic calculations. Soft tissue artifact (STA), the motion of tissue relative to the underlying bones, is a primary cause of error in joint center calculations. A method for mitigating the effects of STA, single-frame optimization (SFO), was introduced and numerically verified in Part 1 of this work, and the purpose of this article (Part 2) is to experimentally compare the results of SFO with a marker-based solution. The experimentation herein employed a single-degree-of-freedom pendulum to simulate human joint motion, and the effects of STA were simulated by affixing the inertial measurement unit to the pendulum indirectly through raw, vacuum-sealed meat. The inertial sensor was outfitted with an optical marker adapter so that its location could be optically determined by a camera-based motion-capture system. During the motion, inertial effects and non-rigid attachment of the inertial sensor caused the simulated STA to manifest via unrestricted motion (six degrees of freedom) relative to the rigid pendulum. The redundant inertial and optical instrumentation allowed a time-varying joint center solution to be determined both by optical markers and by SFO, allowing for comparison. The experimental results suggest that SFO can achieve accuracy comparable to that of state-of-the-art joint center determination methods that use optical skin markers (root mean square error of 7.87–37.86 mm), and that the time variances of the SFO solutions are correlated (r =  0.58–0.99) with the true, time-varying joint center solutions. This suggests that SFO could potentially help to fill a gap in the existing literature by improving the characterization and mitigation of STA in human motion capture.

Effect of shoulder model complexity in upper-body kinematics analysis of the golf swing

The golf swing is a complex full body movement during which the spine and shoulders are highly involved. In order to determine shoulder kinematics during this movement, multibody kinematics optimization (MKO) can be recommended to limit the effect of the soft tissue artifact and to avoid joint dislocations or bone penetration in reconstructed kinematics. Classically, in golf biomechanics research, the shoulder is represented by a 3 degrees-of-freedom model representing the glenohumeral joint. More complex and physiological models are already provided in the scientific literature. Particularly, the model used in this study was a full body model and also described motions of clavicles and scapulae. This study aimed at quantifying the effect of utilizing a more complex and physiological shoulder model when studying the golf swing. Results obtained on 20 golfers showed that a more complex and physiologically-accurate model can more efficiently track experimental markers, which resulted in differences in joint kinematics. Hence, the model with 3 degrees-of-freedom between the humerus and the thorax may be inadequate when combined with MKO and a more physiological model would be beneficial. Finally, results would also be improved through a subject-specific approach for the determination of the segment lengths.

Joint Center Estimation Using Single-Frame Optimization: Part 1: Numerical Simulation

The biomechanical models used to refine and stabilize motion capture processes are almost invariably driven by joint center estimates, and any errors in joint center calculation carry over and can be compounded when calculating joint kinematics. Unfortunately, accurate determination of joint centers is a complex task, primarily due to measurements being contaminated by soft-tissue artifact (STA). This paper proposes a novel approach to joint center estimation implemented via sequential application of single-frame optimization (SFO). First, the method minimizes the variance of individual time frames’ joint center estimations via the developed variance minimization method to obtain accurate overall initial conditions. These initial conditions are used to stabilize an optimization-based linearization of human motion that determines a time-varying joint center estimation. In this manner, the complex and nonlinear behavior of human motion contaminated by STA can be captured as a continuous series of unique rigid-body realizations without requiring a complex analytical model to describe the behavior of STA. This article intends to offer proof of concept, and the presented method must be further developed before it can be reasonably applied to human motion. Numerical simulations were introduced to verify and substantiate the efficacy of the proposed methodology. When directly compared with a state-of-the-art inertial method, SFO reduced the error due to soft-tissue artifact in all cases by more than 45%. Instead of producing a single vector value to describe the joint center location during a motion capture trial as existing methods often do, the proposed method produced time-varying solutions that were highly correlated (r > 0.82) with the true, time-varying joint center solution.