Can the Effect of Soft Tissue Artifact Be Eliminated in Upper-Arm Internal-External Rotation?

Transcription

1 Journal of Applied Biomechanics, 2011, 27, Human Kinetics, Inc. Can the Effect of Soft Tissue Artifact Be Eliminated in Upper-Arm Internal-External Rotation? Yanxin Zhang, David G. Lloyd, Amity C. Campbell, and Jacqueline A. Alderson The purpose of this study was to quantify the effect of soft tissue artifact during three-dimensional motion capture and assess the effectiveness of an optimization method to reduce this effect. Four subjects were captured performing upper-arm internal-external rotation with retro-reflective marker sets attached to their upper extremities. A mechanical arm, with the same marker set attached, replicated the tasks human subjects performed. Artificial sinusoidal noise was then added to the recorded mechanical arm data to simulate soft tissue artifact. All data were processed by an optimization model. The result from both human and mechanical arm kinematic data demonstrates that soft tissue artifact can be reduced by an optimization model, although this error cannot be successfully eliminated. The soft tissue artifact from human subjects and the simulated soft tissue artifact from artificial sinusoidal noise were demonstrated to be considerably different. It was therefore concluded that the kinematic noise caused by skin movement artifact during upper-arm internal-external rotation does not follow a sinusoidal pattern and cannot be effectively eliminated by an optimization model. Keywords: biomechanics, kinematics, modeling, motion analysis, optimization Internal rotation of the upper arm is critical to a wide range of daily human and sports activities. In anatomical, sports, and clinical practices, in vivo quantification of the internal rotation angles is an important measurement to assess function (Cheng et al., 2000; Doorenbosch et al., 2003; Elliott et al., 1995). Three-dimensional (3-D) stereophotogrammetric analyses are the current gold standard for unconstrained quantification of joint kinematics. These systems rely on motion cameras and marker sets attached to the skin. Arguably, the greatest limitation of this approach is the soft tissue artifact that results in miscellaneous movements of skin-attached marker sets, relative to the underlying skeletal motion. This artifact can result in substantial errors in the calculation of joint angles, including upper-arm internal rotation (Cutti et al., 2005). Unlike high-frequency random noise, skin-fixed markers move relative to bone in a smooth, continuous way (Cappozzo et al., 1993) but the pattern of soft tissue artifact error has been found to be task dependent and has no distinct region of the low-frequency domain (Fuller et al., 1997). Therefore, traditional filtering techniques cannot successfully remove soft tissue artifact error. Yanxin Zhang (Corresponding Author) is with the Department of Sport and Exercise Science, University of Auckland, Auckland, New Zealand. David G. Lloyd is now with Musculoskeletal Research, Griffith University, Gold Coast Campus, QLD, Australia. Amity C. Campbell is with the School of Physiotherapy, Curtin University of Technology, Bentley, Perth, WA, Australia. Jacqueline A. Alderson is with the School of Sport Science, Exercise, and Health, University of Western Australia, Crawley, WA, Australia. Soft tissue artifact can be reduced for certain movements, such as knee flexion/extension, by carefully locating markers on body surface areas with low soft tissue artifact. In addition, Cappello et al. (1997) recommended a double-calibration method to reduce soft tissue artifact, requiring two static calibration trials to define anatomical landmarks at the two extremes of the expected range of motion. However, this method must be designed specifically for the motor task under analysis, restricting its use as a generic method. Therefore, for joint rotations with large amounts of soft tissue artifact (e.g., upper-arm internal-external rotation), more sophisticated techniques are required. More recently, mathematic models based on optimization theory have been proposed to eliminate soft tissue artifact error. In a mathematical sense, a large group of these models minimize marker array deformation from its reference shape in a least-squares sense (Challis, 1995; Lu & O Connor, 1999; Spoor & Veldpaus, 1980). Simulation studies (Challis, 1995; Lu & O Connor, 1999) showed that optimization models can effectively eliminate artificial noise added to experimental data. Artificial continuous noise of a sinusoidal form has been used to simulate soft tissue artifact (Cheze et al., 1995; Lu & O Connor, 1999), as it provides the flexibility to examine soft tissue artifact of differing amplitude and frequency. However, in a number of instances, this approach may be questionable, as the pattern of real soft tissue artifact is still largely unknown. Therefore, further validation investigations, that include both artificial and real soft tissue artifact noise, are required. 258

2 Skin Movement Artifact 259 In a recent study, Stagni et al. (2009) assessed the performance of global optimization in reducing the affects of soft tissue artifact during the measurement of knee joint kinematics, while concurrently recording 3-D fluoroscopy for use as the gold standard comparison. Their mean root mean square error (RMSE) was up to 25 for joint angles following optimization. This finding was in contrast to the simulation results of Lu and O Connor (1999) in which the RMSEs were less than 3. These contradictory results further highlight the need to carry out validation investigations that do not rely on only in vivo studies or simulation studies. To further investigate soft tissue artifact, this study aimed to test the performance of the optimization model for upper-arm internal-external rotation movements on human subjects and on a mechanical arm with additional artificial sinusoidal noise, as this type of noise has typically been used previously (Cheze et al., 1995; Lu & O Connor, 1999). It was hypothesized that (1) soft tissue artifact follows a sinusoid-like motion pattern and (2) the optimization model can reduce soft tissue artifact for upper-arm internal rotations. Methods The data collection was carried out in two parts. In the first part, four male human subjects (age range: 20 22) were recruited and tested at the University of Auckland to quantify the effect of real soft tissue artifact during upper-arm internal-external rotation. Each subject provided informed consent following the study s ethical approval from the University of Auckland s Human Research Ethics Committee. The second part of the data collection was carried out using a mechanical arm to simulate upper-arm internal-external rotations without soft tissue artifact. This data were collected at the University of Western Australia. In both sets of experiments, a Vicon MX motion analysis system operating at 250 Hz was used to record the motion of surface markers that were attached to the human subjects and to the mechanical arm. Procedures Before recording the actual motion trials, anatomical landmark calibration trials were performed. In these trials, subjects were required to maintain a static posture with the elbow in 90 flexion allowing all marker positions to be recorded. Upper-body markers included the seventh cervical vertebra (C7), 10th thoracic vertebra (T10), sternoclavicular notch (CLAV), xyphoid process of the sternum (STRN), posterior shoulder, anterior shoulder, elbow medial epicondyle (EM), elbow lateral epicondyle (EL), most caudal-lateral point on the radial styloid (RS), caudal-medial point on the ulnar styloid (US), a triad of markers affixed to the proximal upper arm, and a triad of markers affixed to the distal upper arm. An illustration of the marker set is shown in Figure 1, and the abbreviations used in the figures are given in Table 1. Proximal upperarm and distal upper-arm triads were positioned in areas that were not largely influenced by the soft tissue artifact according to Campbell et al. (2009a, 2009b). Medial and lateral elbow epicondyle markers were removed for the dynamic movement trials, which entailed each subject rotating their upper arm 90 internally and then externally, three to five times. Throughout the motion trials, the upper arm was abducted to shoulder level (90 relative to the thorax), and the elbow was held in 90 of flexion (Figure 2). The subjects held their forearms in a neutral pronation supination posture and were instructed to not pronate or supinate their forearm during the shoulder internal-external rotation movements. Table 1 Key to the abbreviations used in the figures Abbreviation ASH C7 CLAV EC EL EM PSH RS SC STRN T10 US Anatomical Location Anterior Shoulder 7th Cervical Vertebra Clavicular Notch Elbow Centre Lateral Epicondyle Medial Epicondyle Posterior Shoulder Radial Styloid Shoulder Center Sternum 10th Thoracic Vertebra Ulnar Styloid Figure 1 Illustration of the upper body marker set. See Table 1 for a key to the abbreviations.

3 260 Zhang et al. determined as a point located along the vector formed by the ulnar styloid and radial styloid markers at a distance of 7.5 mm (marker radius) toward the radial styloid from the ulnar styloid marker. Following the recommendations of the International Society of Biomechanics, two different upper-arm coordinate systems were defined (Table 2) based on (1) upper-arm anatomical landmarks that comprised the shoulder joint center and elbow markers (UACS EL ) and (2) the anatomical landmarks of both the upper arm and forearm (UACS FA ). The UACS FA was based on the long axes of the upper arm and forearm (Table 2). Due to the short distance between the medial and lateral elbow epicondyle markers, UACS EL had the potential to have a high level of measurement error due to skin movement (Wu Figure 2 Experiment setup with a subject in the testing posture. The mechanical arm was designed and constructed to represent the upper-limb segments, including a replica shoulder and elbow joint (Elliott et al., 2007). The same landmarks and markers were mounted on the surface of the segments of the mechanical arm (shown later in Figure 5). The mechanical arm was manually manipulated to replicate the same 90 internal-external rotation trials as those described for the human subject. Given that the mechanical arm did not suffer from miscellaneous marker movement, these trials and the subsequent analyses served as movement trials without soft tissue artifact. Coordinate System Definition and Marker Locations A biomechanical model was developed in this study based on a previous upper-limb model (Lloyd et al., 2000; Campbell et al., 2009a, 2009b). Three rigid body segments were defined thorax, upper arm, and forearm based on the anatomical landmark positions. The calibrated anatomical systems technique (Cappozzo et al., 1995) was used to establish the motion of virtual anatomical landmarks relative to upper-arm and forearm technical coordinate systems. The virtual positions of the upper-limb landmarks (medial and lateral elbow epicondyle markers) were determined relative to the upper-arm technical coordinate system, which was defined using the proximal upper-arm triad. The shoulder joint center was determined as midway between the posterior and anterior shoulder markers and the elbow joint center was determined as the midpoint between the medial and lateral elbow epicondyle markers (Figures 1, 3, and 4). The motion of these virtual upper-limb landmarks was then reconstructed from their constant relative positions to the upper-arm technical coordinate system. For the forearm, the motion of the ulnar styloid process was Figure 3 Illustration of the UACS EL and torso coordinate systems. Figure 4 Illustration of the UACS FA and torso coordinate systems (y f is the long axis vector of the forearm).

4 Skin Movement Artifact 261 Table 2 Definitions of the upper-arm and torso anatomical segment coordinate systems Name Definition Torso Origin: C7. Y: Unit vector going from T10 to C7. Z: Unit vector perpendicular to the sagittal plane defined by T10, C7, and CLAV, pointing laterally. X: Unit vector defined by X- and Z-axes to create a right-hand coordinate system. Upper-Arm 1 Origin: The elbow joint center, which was the midpoint between EL and EM. (UACS EL ) X: Unit vector perpendicular to the plane formed by EL, EM, and SC, pointing anteriorly. Y: Unit vector going from the elbow joint center to shoulder joint center. Z: Unit vector defined by X- and Y-axes to create a right-hand coordinate system. Upper-Arm 2 Origin: The elbow joint center, which was the midpoint between EL and EM. (UACS FA ) X: Unit vector perpendicular to the Z- and Y-axes, pointing anteriorly. Y: Unit vector going from the elbow joint center to shoulder joint center. Z: Unit vector perpendicular to the plane formed by Y-axis of the upper arm and the long axis vector of the forearm. et al., 2005). The International Society of Biomechanics recommended using UACS FA when the forearm is available from the recordings. In this study, we used internalexternal rotation angles calculated using the UACS FA as the gold standard to evaluate the soft tissue artifact in using the UACS EL. It should be noted that defining the UACS FA is only possible when the long axes of the upper arm and forearm are not close to being parallel, which was ensured in our motion analysis testing methods. Optimization Model An optimization model (Challis, 1995; Lu & O Connor, 1999) was applied to the kinematic data to minimize soft tissue artifact. Marker positions from the static calibration trial were free from soft tissue artifact and were taken as reference input into the model. The transformation matrix from the global coordinate system and anatomical coordinate system was therefore initially established. It needs to be noted that the accuracy of the rigid body transformation parameters are influenced by marker numbers. As the number of markers increased, the accuracy of the estimation of the transformation parameters increased, and the increase was most rapid when the number of markers was increased from three to four (Challis, 1995). However, a greater number of markers may also result in tracking problems (i.e., errors in the motion analysis system reconstruction) and interfere with the execution of dynamic movements. In this study, the position of the six markers, comprising the proximal upper-arm and distal upper-arm triads relative to the UACS EL were determined. During the dynamic trials, due to soft tissue artifact, the relative position of the markers in the anatomical coordinate system changes. To compensate for this effect, the optimization process was conducted by minimizing the weighted sum of squared distances between simulated and model-determined marker positions. In this study, soft tissue artifact for every marker was equally weighted and a singular value decomposition algorithm (Challis, 1995) was used to compute the optimal transformation parameters, which minimized the transformation variances. Data Analysis Based on the recorded markers positions, the anatomical coordinate systems were defined according to the biomechanical model. Euler angles were used to calculate the upper-arm internal-external rotation angles with respect to the torso reference segment by Y-X-Y rotation sequence. This sequence follows the International Society of Biomechanics recommendation with the rotation order of plane of elevation (rotation about the Y-axis in the torso), elevation angle (rotation about the X-axis that is in the torso transverse plane but perpendicular to the plane of elevation), and internal/external rotation (rotation about the Y-axis of the upper arm) (Wu et al., 2005). The Euler angles were processed in different ways for human data and mechanical arm data. For human data, the gold standard shoulder internal-external rotation angles were calculated based on the transformation matrix between UACS FA and torso coordinate systems as per the International Society of Biomechanics recommendation since the long axis of the forearm, used to define the UACS FA, well represents upper-arm internal-external rotation and is less affected by soft tissue artifact (Wu et al., 2005). The internalexternal rotation angles were also calculated based on the transformation matrix between UACS EL and torso coordinate systems, which would be affected by soft tissue artifact. This is because the medial and lateral elbow epicondyle marker positions used in defining UACS EL were reconstructed from the motion of the upper-arm technical CS that was defined from the proximal upper arm triad. Subsequently, angles determined using the UACS EL were termed the noisy internal-external rotation

5 262 Zhang et al. angles. The optimization model was applied to eliminate soft tissue artifact and modify noisy internal-external rotation angles. It should be noted that an angular offset between noisy and gold standard internal-external rotation angles exists, due to the different definitions of the upper-arm coordinate systems (Cutti et al., 2005). In this study, the angular offset was determined from the calibration trial and added to the gold standard internal-external rotation angles to ensure that two angles had the same starting position. Finally, the noisy internal-external rotation angles, with and without optimization, were compared with the gold standard angles, and the RMSE calculated. The average and the standard deviation of the RMSE for all trials were computed. A two-tailed t test was performed to compare the RMSE values under two conditions (with and without optimization). To account for multiple comparisons, the confidence interval was set to be 99% (p =.01). For the mechanical arm data, internal-external rotation angles were calculated based on the transformation matrix between the UACS EL and torso coordinate systems. Since there is no soft tissue artifact in the mechanical data, this angle was set as the gold standard. Sinusoidal kinematic noise signals (Cheze et al., 1995; Lu & O Connor, 1999) with 5 mm, 10 mm, 15 mm, 20 mm, and 25 mm amplitudes were added to proximal upperarm and distal upper-arm position data for each trial, to simulate soft tissue artifact of the upper-arm segment. The optimization model was then applied to the noisy data. The internal-external angles with and without optimization were derived, compared with the gold standard data, and the RMSE calculated. The average and standard deviation of the RMSE for all trials under each noise level were computed. A two-tailed t test was performed to compare the RMSE values under two conditions (with and without optimization). Discussion The purpose of this investigation was to quantify soft tissue artifact during 3-D motion capture and demonstrate the in vivo validity of a soft tissue artifact minimization technique: the optimization method. It was hypothesized that (1) the optimization model can reduce soft tissue artifact for upper-arm internal-external rotation, and (2) soft tissue artifact follows a sinusoid-like motion pattern. The results from both human and mechanical arm kinematic data demonstrated that the soft tissue artifact can be reduced by the optimization model. However, for the human data, the 17 mean RMSE following optimization was still very high. For the mechanical arm data, the largest RMSE was 14 with the sinusoidal noise level of 25 mm and a mean RMSE of 8. It requires noting that the noise level of 25 mm is higher than the normal range of soft tissue artifact (Cappozzo et al., 1993). However, even for this extreme case, the sinusoidal pattern of noise was well removed in the mechanical arm kinematic data, with an RMSE value considerably lower than the values from obtained from the human data. We therefore conclude that Results Mechanical Arm Data There was a noticeable difference between the two upperarm internal-external rotation angles computed by noisy internal-external rotation angles and optimization (Figure 8). For each level of noise, the mean RMSE value for optimization was lower than the noisy internal-external rotation angles. However, large variances were observed and therefore the results were not significantly different (p >.01). Increased amplitudes of noise also caused increases of RMSE values from 1 to over 8. Human Subject Data The results demonstrate that there was a significant difference (p <.01) between the internal-external rotation angles computed by noisy internal-external rotation angles and optimization (Figures 6 and 7). The RMSE for noisy internal-external rotation angles was 24 ± 1.9 and the RMSE for optimization was decreased to 17 ± 1.3. Figure 5 The mechanical arm with attached upper-limb markers (only the labeled markers were used; the torso, distal upper arm, and PSH markers are not shown in the picture).

6 Figure 6 Typical of angular trajectories of the upper arm internal-external rotation angle for the human subject (dashed curve = with optimization; dotted curve = without optimization; solid curve = gold standard). Figure 7 Mean/standard deviation of RMSE for human data. NIER = noisy internal-external rotation. Figure 8 Mean/standard deviation of RMSE for mechanical arm data. NIER = noisy internal-external rotation 263

7 264 Zhang et al. the soft tissue artifact for upper-arm internal rotations is probably not well represented by a sinusoidal motion pattern. Consequently, the second hypothesis was rejected. The results from simulation (Challis, 1995; Lu & O Connor, 1999) indicate that the optimization model can efficiently eliminate artificially generated noise by singular value decomposition. However, our results indicate that soft tissue artifact for upper-arm internalexternal rotation has more continuous, systematic patterns that cannot be successfully eliminated through an optimization model. There a number of possible limitations in our study. Carrying out experiments on four male subjects of the same form may limit the generalization of the results. However, these subjects would provide the best performing data for the model and motion analysis methods as these would have the least amount of soft tissue artifact. Other groups with more adipose tissue and skin would probably exhibit worse results. Our research was limited to soft tissue artifact that may occur in upper-arm internal-external rotation. However, the soft tissue artifact pattern may vary for different segments in different tasks. A single case study is not adequate to conclude that soft tissue artifact cannot be effectively eliminated by an optimization model. However, the results support the argument that systematic general characterization of the soft tissue artifact is not possible, and also not practicable due to large differences among subjects and tasks (Leardini et al., 2005). The results of this investigation suggest that the most realistic way to reduce soft tissue artifact currently is the positioning of markers on areas least affected by soft tissue artifact. Independent of soft tissue artifact, mislocation of shoulder joint and elbow joint centers may cause elbow flexion-extension cross-talk to enter into upper-arm internal-external rotation (Chin et al., 2010). In our study, the upper-arm internal-external rotation axis is the vector connecting elbow joint center to the shoulder joint center. The elbow joint center is an appropriate distal end of the upper arm and should not move relative with internalexternal rotation. As we have reported, mislocation of the elbow joint center is probably only small when determined by the location of epicondyles (Chin et al., 2010). The upper arm rotates about the shoulder joint center; hence, defining the proximal end of the upper arm in the shoulder joint center is appropriate. However, we have reported a mislocation of the shoulder center (Campbell et al., 2009a, 2009b) of about 1 cm when using anterior and posterior markers on the shoulder. We simulated this effect by deviating the real shoulder center position of the robot arm in a 1 cm cube. We found the effect of this result in an internal-external angle of less than 1. Compared with the 24 caused by soft tissue artifact, the effect of shoulder joint center deviation is not a major concern in the calculation of internal-external rotation angles. As the gold standard in this study, we use internalexternal rotation angles calculated using the UACS FA, which are believed to be minimally affected by soft tissue artifact (Wu et al., 2005). However, for a large number of sports activities that include throwing, such as cricket bowling and tennis serving, the internal rotation of the upper arm is associated with the forearm fully or close to full extension at some phase of the task (Elliott et al., 1995, 2007). Given that the upper arm and forearm long axes are collinear at some point during the execution of these dynamic tasks, the calculation of internal-external rotation angles using the UACS FA will generate erroneous measures. This method therefore has limited applicability to tasks that maintain forearm flexion throughout. Further confounding the application of this method is any magnitude of forearm pronation-supination, which will alter the position of the wrist markers, and in turn result in errors in the calculation of the upper-arm s internal-external rotation angle. Although Cutti et al. (2005) developed a compensation algorithm to eliminate this affect, their method still cannot be applied when collinearity or near collinearity of the forearm and upper arm exists. In these cases one must rely on the internal-external rotation angles calculated using the UACS EL, which we have shown is prone to soft tissue artifact that cannot be reduced using optimization methods. As indicated by the results of this investigation, new methods need to be developed to compensate for soft tissue artifact of the upper arm so that internal-external rotation can be measured with the elbow close to or at full extension. This correction could be a cross-calibration between the different methods used to determine upper-arm internalexternal rotation, similar to the data shown in Figure 6. However, this would have to be developed and verified to work across a range of different types of subjects and movements. This is beyond the scope of the current paper, although we are currently developing and testing an online soft tissue artifact compensation algorithm using both upper-arm markers and forearm markers. Acknowledgments The support of Vicon Peak Motion Systems in carrying out this work is greatly appreciated. References Campbell, A.C., Alderson, J.A., Lloyd, D.G., & Elliott, B.C. (2009a). Effects of different technical coordinate system definitions on the three dimensional representation of the glenohumeral joint centre. Medical & Biological Engineering & Computing, 47(5), Campbell, A.C., Lloyd, D.G., Alderson, J.A., & Elliott, B.C. (2009b). MRI development and validation of two new predictive methods of glenohumeral joint centre location identification and comparison with established techniques. Journal of Biomechanics, 42, Cappello, A., Cappozzo, A., La Palombara, P.F., Lucchetti, L., & Leardini, A. (1997). Multiple anatomical landmark calibration for optimal bone pose estimation. Human Movement Science, 16(2 3), Cappozzo, A., Catani, F., & Leardini, A. (1993) Skin movement artifacts in human movement photogrammetry. Proceed-

8 Skin Movement Artifact 265 ings of the XIV th Congress of the International Society of Biomechanics, Vol. I, pp Paris, France. Cappozzo, A., Catani, F., Della Croce, U., & Leardini, A. (1995). Position and orientation in space of bones during movement: anatomical frame definition and determination. Clinical Biomechanics (Bristol, Avon), 10, Challis, J.H. (1995). A procedure for determining rigid body transformation parameters. Journal of Biomechanics, 28(6), Cheng, P., Nicol, A., & Paul, J. (2000). Determination of axial rotation angles of limb segments a new method. Journal of Biomechanics, 33, Cheze, L., Fregly, B.J., & Dimnet, J. (1995). A solidification procedure to facilitate kinematic analyses based on video system data. Journal of Biomechanics, 28(7), Chin, A., Lloyd, D., Alderson J., Elliott B., & Mills, P. (2010). A marker-based mean finite helical axis model to determine elbow rotation axes and kinematics in vivo, Journal of Applied Biomechanics, 26, Cutti, A.G., Paolini, G., Troncossi, M., Cappello, A., & Davalli, A. (2005). Soft tissue artefact assessment in humeral axial rotation. Gait & Posture, 21, Doorenbosch, C.A.M., Veeger, H.E.J., & Harlaar, J. (2003). The globe system: An unambiguous description of shoulder positions in daily life movements. Journal of Rehabilitation Research and Development, 40, Elliott, B.C., Alderson, J.A., & Denver, E.R. (2007). System and modelling errors in motion analysis: implications for the measurement of the elbow angle in cricket bowling. Journal of Biomechanics, 40(12), Elliott, B., Marshall, R., & Noffal, G. (1995). Contributions of upper limb segment rotations during the power serve in tennis. Journal of Applied Biomechanics, 11, Fuller, J., Liu, L.J., Murphy, M.C., & Mann, R.W. (1997). A comparison of lower-extremity skeletal kinematics measured using skin- and pin-mounted markers. Human Movement Science, 16(2-3), Leardini, A., Chiari, L., Della Croce, U., & Cappozzo, A. (2005). Human movement analysis using stereophotogrammetry. Part 3. Soft tissue artifact assessment and compensation. Gait & Posture, 21(2), Lloyd, D.G., Alderson, J., & Elliott, B.C. (2000). An upper limb kinematic model for the examination of cricket bowling: a case study of Mutiah Muralitharan. Journal of Sports Sciences, 18(12), Lu, T.W., & O Connor, J.J. (1999). Bone position estimation from skin marker co-ordinates using global optimisation with joint constraints. Journal of Biomechanics, 32(2), Spoor, C.W., & Veldpaus, F.E. (1980). Rigid body motion calculated from spatial co-ordinates of markers. Journal of Biomechanics, 13(4), Stagni, R., Fantozzi, S., & Cappello, A. (2009). Double calibration vs. global optimisation: performance and effectiveness for clinical application. Gait & Posture, 29(1), Wang, X., Maurin, M., Mazet, F., De Castro Maia, N., Voinot, K., Verriest, J.P., et al. (1998). Three-dimensional modelling of the motion range of axial rotation of the upper-arm. Journal of Biomechanics, 31, Wu, G., van der Helm, F.C.T., Veeger, H.E.J., Makhsous, M., van Roy, P., Anglin, C., Nagels, J., Karduna, A.R., McQuade, K., Wang, X., Werner, F.W., & Buchholz, B. (2005). ISB recommendation on definitions of joint coordinate systems of various joints for the reporting of human joint motion Part II: shoulder, elbow, wrist and hand. Journal of Biomechanics, 38(5),