- Email: [email protected]
A framework for the definition of standardized protocols for measuring upper-extremity kinematics
Clinical Biomechanics 24 (2009) 246–253
Contents lists available at ScienceDirect
Clinical Biomechanics journal homepage: www.elsevier.com/locate/clinbiomech
Review
A framework for the definition of standardized protocols for measuring upper-extremity kinematics A. Kontaxis a, A.G. Cutti b, G.R. Johnson a, H.E.J. Veeger c,d,* a
Newcastle University, Newcastle upon Tyne, United Kingdom INAIL Prostheses Centre, Vigorso di Budrio, Italy c VU University Amsterdam, Amsterdam, The Netherlands d Delft University of Technology, Delft, The Netherlands b
a r t i c l e
i n f o
Article history: Received 23 July 2008 Accepted 24 December 2008
Keywords: Shoulder Elbow Kinematics Framework Standard
a b s t r a c t Background: Increasing interest in upper extremity biomechanics has led to closer investigations of both segment movements and detailed joint motion. Unfortunately, conceptual and practical differences in the motion analysis protocols used up to date reduce compatibility for post data and cross validation analysis and so weaken the body of knowledge. This difficulty highlights a need for standardised protocols, each addressing a set of questions of comparable content. The aim of this work is therefore to open a discussion and propose a flexible framework to support: (1) the definition of standardised protocols, (2) a standardised description of these protocols, and (3) the formulation of general recommendations. Methods: Proposal of a framework for the definition of standardized protocols. Findings: The framework is composed by two nested flowcharts. The first defines what a motion analysis protocol is by pointing out its role in a motion analysis study. The second flowchart describes the steps to build a protocol, which requires decisions on the joints or segments to be investigated and the description of their mechanical equivalent model, the definition of the anatomical or functional coordinate frames, the choice of marker or sensor configuration and the validity of their use, the definition of the activities to be measured and the refinements that can be applied to the final measurements. Finally, general recommendations are proposed for each of the steps based on the current literature, and open issues are highlighted for future investigation and standardisation. Interpretation: Standardisation of motion analysis protocols is urgent. The proposed framework can guide this process through the rationalisation of the approach. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction In the past 20 years, quantitative motion analysis techniques have provided answers to a wide variety of problems concerning upper-extremity biomechanics. Moreover, different authors have addressed similar problems from different perspectives (Anglin and Wyss, 2000b). At this stage of development, one would expect, therefore, to be able to compare or integrate the results from different studies to draw stronger or broader conclusions. Unfortunately, this is not always possible, due to conceptual and practical differences in the motion analysis protocols used resulting in the application of different techniques or terminologies to define similar measurable quantities (Anglin and Wyss, 2000b). For instance, different authors used different kinematic models of the upper-limb (van der Helm, 1994b; Murray and Johnson, * Corresponding author. Address: VU University Amsterdam, Amsterdam, The Netherlands. E-mail address: [email protected] (H.E.J. Veeger). 0268-0033/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.clinbiomech.2008.12.009
2004), coordinate systems (Anglin and Wyss, 2000a; Cutti et al., 2008), motor tasks (Murray and Johnson, 2004, van Andel et al., 2008), scapula or clavicle tracking or optoelectronic marker setups (Johnson et al., 1993; Karduna et al, 2001; Ludewig et al., 2004; Anglin and Wyss, 2000a). This difficulty highlights a need for standardised protocols for motion analysis of the upper extremity. Although it is almost impossible to define a single, universal protocol able to address all possible questions, at least a certain degree of standardisation should be sought. In particular, it should be possible to agree on general recommendations and to define standardised protocols, each addressing a set of questions of comparable content. The aim of this work is therefore to open a discussion and propose a flexible framework to support: (1) the definition of standardised protocols, (2) a standardised description of these protocols, and (3) the formulation of general recommendations. Through the framework, the authors aim to clear confusion regarding kinematics results obtained from anatomical and functional frames, as well as giving recommendations on how to use these
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
different coordinate systems. Finally, the study aims to expose the lack of knowledge in some areas of upper extremity kinematics where there is poor or no standardisation and so trigger further investigations that will bridge the gap. 2. Methods The proposed framework consists of two nested flowcharts. The former (Section 2.1, Fig. 1) defines what is meant by a motion analysis protocol by pointing out its role in a measurement study. The latter (Section 2.2, Fig. 2a) specifies the steps required to build such a motion analysis protocol, appropriate to the study design. Basic recommendations and current gaps for issues discussed in the literature are also discussed in Section 3 and presented in Fig. 2b. 2.1. Definition of a motion analysis protocol As in any scientific investigation, the following basic steps should be taken to successfully conclude a motion analysis study (Fig. 1). First, there is a need to define the (clinical, ergonomic or sports related) research question (Step 1), which in turn defines whether the study should: (i) involve intra- or inter-subject comparisons; (ii) address specific joints, user function assessment, or biomechanical model; (iii) be longitudinal or cross-sectional. A set of hypotheses to be proven or rejected should be formulated in order to address the research question (Step 2), followed by the definition of the parameters required to test the hypotheses (Step 3). Then the Motion Analysis Protocol should be built in order to measure the parameters over the population of the study (Step 4). The additional steps for this purpose are described in Section 2.2 and illustrated in Fig. 2a. After application of this protocol to the population of the study (Step 5) and a statistical analysis over the parameters measured (Step 6) the hypotheses can be verified/falsified (Step 7) to conclude the study. A motion analysis protocol is therefore the means of measurement of the parameters required to test the hypotheses at the base of the study research question. 2.2. Building a motion analysis protocol When constructing a motion analysis protocol the following six issues (Sections 2.2.1–2.2.6, Fig. 2a) must be addressed: 2.2.1. Joint and/or segment kinematics to be studied It is first necessary to specify whether joint and/or segment kinematics is of interest, and to identify the joints and segments involved. Segment kinematics is defined as the attitude of one bony segment with respect to either a global coordinate system or a nonadjacent bone. Examples of segment kinematics are the attitude of the thorax in the global coordinate of the measurement system or the attitude of the humerus relative to the thorax.
247
Joint kinematics is defined as the relative attitude of two adjacent bony segments. Joint kinematics is for example the attitude of the clavicle relative to the thorax, scapula relative to the clavicle, scapula relative to the thorax, or humerus relative to the scapula. As addressed further in Section 2.2.3, the rotational component of joint kinematics can be described through the decomposition of a rotation matrix as a series of rotations around either predefined anatomical axes or joint functional axes (see the Glossary). The choice of anatomical versus functional axes has important implications in the definition of the protocol and in particular in the definition of anatomical/functional coordinate systems. It is important to note that not all joints/segments of interest may be measurable, depending on the instrumentation or time available to perform the measures. For example: (1) recording of the 3D attitude of the clavicle with non invasive methods requires specialised equipment (Marchese and Johnson 2000); (2) scapula tracking for humerus elevation higher than 120°. requires specialised equipment and/or sufficient time (Johnson et al., 1993; Pronk and van der Helm, 1991). The final list of joints or segments of interest should be formulated after considering all measurement and time constraints. 2.2.2. Mechanical model of joints/Degrees of Freedom for segment kinematics If the focus of the study is joint kinematics, a mechanical (kinematic) equivalent of the joints of interest must be formulated, specifying the number of linkages and the Degrees of Freedom (DoFs) (Fig. 3). The mechanical model must take into account known preferential axes of rotation of the joint. On the other hand, if segment kinematics is the object of the study, it should be specified whether segments are considered to have six DoFs (three rotations plus three translations) or otherwise. 2.2.3. Definition of joint or segment coordinate systems and angles Irrespective of whether joint or segment kinematics are to be studied, local coordinate systems are required to define the attitude of one segment relative to another. For the joints of interest, coordinate systems must be defined consistently with the appropriate mechanical models. The definition of a joint coordinate system typically requires the definition of two coordinate systems, one for the proximal and one for the distal segment forming the joint. To ensure optimal comparability between different studies, it is proposed that these two coordinate systems should be anatomical frames (see Glossary and Fig. 4). However, the relative orientation of the anatomical frames of the proximal and distal segment can be only indicative of the actual joint rotations because the anatomical axes forming these frames are generally only a rough approximation of the real joint axes of rotations. The consequence of the approximation is a joint kinematics affected by kinematic cross-talk (Piazza and Cavanagh, 2000). If the calculated rotations have to approximate joint motion closely, or the measuring system precludes the measurement of bony landmarks, functional frames should be used for the
Fig. 1. Flowchart that describes the basic steps to conclude a motion analysis study, thus pointing about the role of a motion analysis protocol.
248
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
Fig. 2. (a) Flowchart that specifies the steps required to build a motion analysis protocol (left panel); (b) basic recommendations and current gaps for each of the steps (right panel). Analytical references are presented in the corresponding sections indicated on the panel.
segments, in which the axes are defined on the basis of functional axes of rotation (see Glossary and Fig. 4). An example of how the choice of anatomical versus functional frames can substantially affect the measurement of elbow kinematics is shown in Appendix (adapted from Cutti et al. (2008)). If two joints are of interest involving a single segment then two coordinate systems may need to be defined for the same segment, one associated with its proximal and one with its distal end. The use of one single segment coordinate system may preclude the possibility to describe appropriately the kinematics of both joints at the two ends of the segment. This, for example, is the case for the humerus and the description of elbow and glenohumeral joints (Cutti et al., 2008) (Fig. 5).
For segment kinematics, segment coordinate systems should be representative of the anatomy of the bone. As for joint kinematics, in order to ensure maximum comparability of results between studies, these should be expressed using anatomical frames (Wu et al., 2005) For interpretation of either joint or segment kinematics, appropriate sequences of Euler angles must be defined to express the relative orientation of coordinate systems in a clinically meaningful manner. 2.2.4. Marker/sensor set-up Once the anatomical/functional coordinate systems are defined, they must be linked to the markers/sensors of the measurement
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
249
Fig. 3. Example of a simple mechanical (kinematic) equivalent for the upper-limb. In (a) the linkages (black lines) and joints (squares) of interest are pointed out, while in (b) the degrees of freedom of each joint are further specified. Acronyms: Elevation–Depression (El–De), Protraction–retraction (Pr–Re), Internal–External rotation (In–Ex), elbow Flexion–Extension (eFl–Ex), Pronation–Supination (Pn–Su), Radio–Ulnar deviation (Rd–Ul), wrist Flexion–Extension (wFl–Ex).
system. This requires (1) to define marker/sensor locations over the subject’s body, and (2) to establish the relationship between the joint/segment coordinate systems and the technical frames (see Glossary) associated with the markers/sensors. 2.2.5. Activities to be measured The protocol should specify the set of activities to be executed in the study and should differentiate between those used for calibration (e.g., set-up of joint coordinate systems, definition of landmarks, joint centres and rotational axes) and the activities aiming to investigate the function. In addition, there must be (1) a full description of the activities to be executed together with (2) a clear description of how they will be presented to the subjects, and (3) a statement of the number of repetitions. 2.2.6. Kinematics estimation and refinement through mechanical/ signal modelling Mechanical models can be applied to estimate or refine the attitude of specific segments, starting from the attitude of other segments or by applying specific constraints over the measured kinematics (Biryukova et al., 2000). Similarly, authors have proposed specific algorithms to compensate for errors affecting segment attitude (Cutti et al., 2005a; Roux et al., 2002; Schmidt et al., 1999). Only methodologies of proven validity should be applied. If refinements are applied the protocol should include a clear description of the methodology applied. 2.3. Current standardisation and issues to be addressed In the following section the authors have tried to summarise, based on the available literature, some basic recommendations and open issues for the six steps required to define a protocol (Section 2.2). A schematic summary of these recommendations and open issues is presented in Fig. 2b. 2.3.1. Joints/segments of interest Depending on the study and the research question the appropriate joints/segments to be measured should be selected. Note that
Fig. 4. Example of Anatomical Frames (AF) for thorax, humerus and forearm, and of Functional Frames (FF) for humerus and forearm to optimally describe the elbow joint kinematics. The AFs are based on the ISB recommendations (Wu et al., 2005) while the FFs are based on the elbow Flexion–Extension (Fl–Ex) and PronoSupination (Pn–Su) axes of rotations (Cutti et al., 2008). Black dashed lines: segment axes; black dashed-dotted lines: elbow joint axes.
250
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
not all joint/segment kinematics may be measurable (e.g. instrumentation or time restrictions), or need to be measured. General recommendations are therefore difficult in this case. However, in functional studies, even if previous publications focused only on shoulder/elbow (Murray and Johnson, 2004; Veeger et al., 2006), hand position and orientation should also be considered since it frequently determines the accomplishment of a functional task (van Andel et al., 2008). This recommendation is supported by the extensive attention given to the hand in clinical scales assessing function and capacity in children (Krumlinde-Sundholm et al., 2007). In some clinical settings, it can be useful to analyse arm motion only, without entering into the details of the scapula or clavicle (Cutti et al., 2005b; Garofalo et al., 2009). 2.3.2. Mechanical model of joints Model definitions vary from linked segment models to complex musculoskeletal models with different DoFs for each joint. For example Murray and Johnson (2004) defined a nine linkage mechanism to describe only functional shoulder, elbow and hand orientations whereas others used more links to measure and describe also the scapula and clavicle (Charlton and Johnson, 2006; van der Helm, 1994a). As a general recommendation, protocols should refer to published models whenever possible. Otherwise, a detailed review of all the DoFs and of the closed or open loop mechanisms should be given for future reference and comparisons between studies. 2.3.3. Definition of joint/segment coordinate systems and angles In combination with the mechanical model definition, consistent coordinate systems should be defined for the joints/segments of interest. It is proposed that the standardisation proposal of the International Society of Biomechanics-International Shoulder Group (ISB-ISG) (Wu et al., 2005) on anatomical frames and rota-
Fig. 6. Appendix figure: description of (i) a technical frame (based on markers M1, M2 and M3), (ii) an anatomical frame for the proximal humerus (based on the position of anatomical landmarks GH, EL, EM) and (iii) a functional frame for the distal humerus (based on the functional Flexion/Extension axis of elbow and on the humerus longitudinal anatomical axis) (a) frontal view; (b) sugittal view.
tion/decomposition orders for the upper extremity should be followed. For the description of glenohumeral rotations, this order implies gimbal lock at 0° and 180° elevation. If motion around these positions is important, the sequence Flexion–Extension, AbAdduction and Internal–External rotation is recommended for movements in the sagittal plane. For frontal movements, the sequence Ab-Adduction, Flexion–Extension and Internal–External rotation is recommended instead (Cutti et al., 2008). For the use of functional frames, no standard yet exists. It is strongly recommended that this standard, when proposed, should include the link to anatomical frames. This could be based upon information regarding the orientation of joint motion axes relative to anatomical coordinate systems (Stokdijk et al., 2000; Veeger et al., 1997) or by direct measurement of both anatomical landmarks and functional axes. It is important to keep in mind that the use of anatomical frames to describe joint rotations will produce apparent joint motion errors, i.e. kinematic cross-talk. If joint motion needs to be estimated using anatomical frames, then corrections for this difference in alignment can be applied (Cutti et al., 2005a). 2.3.4. Marker/sensor set-up The defined coordinate systems must be linked to the sensors traceable by the measurement systems. Anglin and Wyss (2000b) have presented an extended review on marker set-ups and landmark location. However, to the knowledge of the authors, no methodological study is yet available to recommend the optimum set-up for a specific type of pathology or group of subjects, where movement constraints, loose skin and fat issues can create inaccuracies. It should be noted that marker placement over anatomical landmarks can create large skin artefacts (Cappozzo et al., 1996). Thus it is proposed: Not to attach marker/sensor on anatomical landmarks, instead use technical frames (see Glossary), and refer the anatomical/functional frames to the technical frames.
Fig. 5. Definition of two coordinate systems for the humerus (proximal and distal ends) to optimally describe the glenohumeral and elbow rotations. The relative orientation of the humerus proximal and thorax coordinate systems describes the glenohumeral joint rotations; the relative orientation of the forearm and humerus distal coordinate systems describes the elbow joint rotations. The thorax and proximal humerus coordinate systems follow the ISB recommendations (Wu et al., 2005)while the distal humerus and forearm coordinate systems follow (Cutti et al., 2008).
The latter step can be achieved by either: (i) anatomical calibrations to define anatomical landmarks and thus anatomical axes and frames (Cappozzo et al., 2005) (ii) and/or functional movements to define functional axes and thus functional frames (Biryukova et al., 2000; Cutti et al., 2008) – see
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
Glossary, or (iii) the use of a reference pose position with predefined segment orientation in which all joint angles are predefined (Cutti et al., 2008). If reference positions for the arm are used, it is proposed that a position, in which the elbow and forearm are in their mid position, or 90° Flexion and 90° Pronation, should be used. This position is relatively easily adopted, even within patient groups and in the movement mid range, thus ensuring that possible estimation errors are evenly spread. Scapular movement is also an essential component of arm elevation and is frequently compromised in pathologies (Kontaxis and Johnson, 2008, Mell et al., 2005). Tracking of scapula motion has been performed in normal and pathological shoulders, with a static palpation method and this is probably the most validated, but rather slow, method for clinical measurement (Johnson et al., 1993). Recent studies (Karduna et al., 2001; Meskers et al., 2007) have indicated that tracking scapular motion by attaching a sensor on the acromion is valid for humerus elevation up to approximately 120°., but the technique was validated only in young healthy subjects. In pathological shoulders with compromised glenohumeral motion the validity of the method is still uncertain. Therefore, for the time being, a static palpation technique is still recommended. If scapular tracking is not possible (dynamically or statically), the only remaining option for the estimation of scapula orientation is the application of regression equations. These can be constructed on an individual basis, within a particular experiment and application (Veeger et al., 1993), but more generalised regression equations are available (de Groot and Brand, 2001). Optimal results do, however, still require the measurement of the scapula resting position. Few studies that have included the direct measurement of clavicle kinematics, since it is difficult to track axial rotation in-vivo without the use of invasive methods (Ludewig et al., 2004; Marchese and Johnson, 2000; Teece et al., 2008). As an alternative, clavicle kinematics is estimated by minimising either the changes in length of the conoid ligament (Pronk and van der Helm, 1991) or the axial rotations of the stiff AC ligament (Charlton and Johnson, 2006). Any algorithm that estimates either scapula or clavicle kinematics should be fully described because it may influence the final measured results.
2.3.5. Set of activities to be measured Defining the number and set of activities to be measured is challenging and depends strongly on the research question. Regarding the number of activities, joint function studies should be addressed with a limited number of standard activities, e.g. single joint motions in the standard planes. On the other hand, functional studies should consider the type of sport/ergonomic/ orthopaedic pathology examined. A number of user function studies have recommended different sets of activities in order to challenge several aspects of the joints of interest (Anglin and Wyss, 2000a; Murray and Johnson, 2004; Peterson and Palmerud, 1996; van Andel et al., 2008). As a general recommendation, activities included in validated clinical scales should be preferred, (e.g. Constant Shoulder, SF36, Oxford Shoulder, DASH shoulder/Elbow, Mayo Elbow, SHUEE or AHA) in order to enhance the possibility to correlate the results of the motion analysis study with the clinical conclusions. All the standardised protocols should consider the subject’s comfort and disability taking into account mobility or time constraints. The number of repetitions should be optimised depending on the physical ability of the subjects, but should be enough to ensure that the measurements of motion cycles are truly representative. If possible, at least three representative repetitions should be available.
251
For functional activities, instructions should be verbal or visual but not guided by the investigator. Several studies have presented compensation mechanisms and variations in arm kinematics according to pathology (Kontaxis et al., 2007; Veeger et al., 2006) or even adaptation under high performance sport tasks (Borsa et al., 2008). During functional activities, the kinematic pattern can be important to explain pathologies and disabilities. Changes to environmental set-up can affect the measurement. If changes are unavoidable (e.g. everyday objects, furniture) it is recommended to measure a control group for each set-up. Activities that describe calibration or pose set-ups or define joint rotational axes etc., can be guided (physically or with a device) in order to achieve maximum accuracy of the measurements. 2.3.6. Kinematics estimation and refinement through mechanical modelling In shoulder motion analysis, soft tissue artefacts can corrupt the calculation of some joint angles. In particular there are reports that all axial rotations tend to be substantially underestimated (Cutti et al., 2005a; Schmidt et al., 1999) and they should be analysed with great caution. In particular, while intra-subject comparisons may still be possible, inter-subject comparison may not be valid. Compensation techniques have been developed and/or tested on upper-limb kinematics such as the correction of the humeral and forearm axial rotation (Cutti et al., 2005a, 2006a; Roux et al., 2002; Schmidt et al., 1999). However, these techniques are not yet fully or extensively validated for all upper extremity segments. It thus appears to be too early to formulate specific recommendations. Until then, researchers should be aware of the limited validity of some angles, as already pointed out. 2.4. Parameters analysis As an additional part of the framework, the authors believe that the discussion should lead to the formulation of recommendations for the treatment of certain kinematic output/parameters. As stated in its definition, a motion analysis protocol is the means to measure a set of parameters over the population of the study. The parameters to be computed in the most general case may vary from study to study. However, certain types of parameters are recurrent in upper extremity studies. These are typically: angle patterns versus time, e.g. the glenohumeral Flexion– Extension; angle-angle plots, e.g. the scapula medio-lateral rotation vs humerothoracic Flexion; range of motion. At least for these common parameters, the protocol should include a description of: how to present them graphically; and, for angle-time and angleangle patterns, how to normalize and extract confidence bands. Lastly, scapular attitudes have been presented as many different decompositions and relative to thorax, humerus, or global coordinate systems. To ensure maximal compatibility of data, we suggest that scapula attitude should be presented as the scapulo-thoracic attitude, but always in relation to the thoraco-humeral attitude. This will allow for every other transformation. This specification will much improve the communication and comparison of results.
252
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
3. Discussion and conclusions
Appendix A
In this paper the authors have tried to summarise the essential steps to be addressed in the definition of motion analysis protocols focusing on (clinical) upper extremity kinematics. In addition, some basic recommendations have been formulated and open problems have been identified. As stated in the introduction, the aim of this paper was not to define one single standardised protocol, but to provide guidelines for protocol developments, comparable to initiatives elsewhere (Bossuyt et al., 2003). Detailed protocols with specific recommendations will have to follow, to measure parameters useful to answer to questions of comparable content. These detailed standardised protocols should be easily developed following the proposed framework which was specifically developed to be both flexible (to adapt to different studies) and detailed in the content to clearly point out the steps required to define a protocol. Both in the framework and in the recommendations, we have tried to point out the difference between segment and joint kinematics, as well as the difference between anatomical and functional frames. This effort was motivated by the feeling of confusion regarding the meaning of the angle patterns obtained by decomposing the relative orientation of the anatomical frames of two adjacent segments. As already pointed out, the axes of anatomical frames are only rough approximations of the real joint axes of rotations. As a consequence, the angles may not closely follow the actual rotations of the joint. Only rotations obtained by decomposing the orientation around functional axes can give indications of the real joint rotations. The confusion that sometimes emerges can also be due to the terminology originally proposed in the ISB recommendations for ‘‘Joint Coordinate Systems”. We think that efforts should be put in reviewing the terminology to enhance clarity. This appears even more urgent since the importance and the attention towards the estimation of functional frames will most probably increase in the next future, due to the diffusion of new measurement systems based in inertial sensors: these systems, in fact, are not (yet) able to measure the position of single anatomical landmarks, but only orientation. By pointing out the essential steps for the definition of a protocol, we hope also to have clarified that a protocol definition is not merely related to ‘‘where to position sensors or markers on the body”, but a much more complex task based on the definition of a kinematic model of the study objective. Finally, we would like to identify the questions that, in the future, will require to be further discussed and addressed:
A.1. Building a technical, an anatomical, and a functional frame: application to the upper arm and to the Humerus (Fig. 6)
(1) the clear distinction between joint angles obtained from anatomical and functional frames; (2) the formulation of guidelines for scapula tracking; (3) the comparison of different set-ups for motion analysis; (4) the problem of soft tissue artefact compensation.
A.1.1. Describing a technical frame on the upper arm Given three markers M1, M2 and M3 attached on the skin of the upper arm, the TF can be constructed as follows:
Y TF ¼ M 2 M1 =kM1 M2 k Z TF ¼ Y TF ^ ðM1 M3 Þ=kX TF ^ ðM1 M 3 Þk X TF ¼ Y TF Z THX The above technical frame has an arbitrary alignment and it could have been defined using a different geometric rule. The markers though are placed in a high visibility position in order to maximise traceability. A.1.2. Humeral anatomical frame (proximal) Following the ISB recommendations (Wu et al., 2005) the anatomical frame H1 of the humerus is constructed based on the centre of rotation of the glenohumeral joint (GH) and the medial (EM) and lateral epicondyles (EM):
Y H1 ¼ ðGH EÞ=kðGH EÞk : longitudinal Z H1 ¼ Y H1 ^ ðEM ELÞ=kY H1 ^ ðEM ELÞk : antero-posterior X H1 ¼ ðY H1 ^ Z H1 Þ=k k : medio-lateral E ¼ ðEL þ EMÞ=2 GH ¼ Origin of the frame The axes of H1 can be used to describe both the glenohumeral joint rotations as well as the elbow joint rotations. While reasonable for the description of the GH joint kinematics, the use of Humeral AF (H1) for the elbow leads to a consisted kinematic cross-talk, which materializes in apparent rotations of the forearm, during pure Flexion–Extensions of the elbow (Cutti et al., 2006b) A.1.3. Humeral functional frame (distal) Following Cutti et al. (2008) the functional frame for the humerus, intended to describe the elbow Flexion–Extension, is constructed as follows:
X HD ¼ V FLEX =kV FLEX k : lateral Z HD ¼ X HD ^ Y H1 =kX HD ^ Y H1 k : posterior Y HD ¼ Z HD ^ X HD =kZ HD ^ X HD k : cranial where VFLEX is the mean Flexion–Extension axis of rotation of the elbow computed through the helical axis algorithm (Veeger et al., 1997), while YH1 is the longitudinal anatomical axis of the humerus (see above). The use of this FF compared to the anatomical frame H1 minimises the kinematic cross-talk at the elbow joint. References
Finally, the authors would reiterate that this paper should be seen as the basis for discussion in the research and clinical communities and eventually lead to an agreed set of protocols and standards.
Acknowledgments The authors gratefully acknowledge the support of Michele Raggi, BEng, INAIL Prostheses Centre, in the preparation of the illustrations. The authors would like to acknowledge the significant contribution to this aspect shoulder biomechanics made by the late Ingram Murray who died at the tragically young age of 35 in December 2008.
Anglin, C., Wyss, U.P., 2000a. Arm motion and load analysis of sit-to-stand, standto-sit, cane walking and lifting. Clin. Biomech. 15 (6), 441–448. Anglin, C., Wyss, U.P., 2000b. Review of arm motion analyses. Proc. Inst. Mech. Eng. [H.] 214 (5), 541–555. Biryukova, E.V., Roby-Brami, A., Frolov, A.A., Mokhtari, M., 2000. Kinematics of human arm reconstructed from spatial tracking system recordings. J. Biomech. 33 (8), 985–995. Borsa, P.A., Laudner, K.G., Sauers, E.L., 2008. Mobility and stability adaptations in the shoulder of the overhead athlete: a theoretical and evidence-based perspective. Sports Med. 38 (1), 17–36. Bossuyt, P.M., Reitsma, J.B., Bruns, D.E., Gatsonis, C.A., Glasziou, P.P., Irwig, L.M., Lijmer, J.G., Moher, D., Rennie, D., de Vet, H.C., 2003. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ 326 (7379), 41–44. Cappozzo, A., Catani, F., Leardini, A., Benedetti, M.G., Croce, U.D., 1996. Position and orientation in space of bones during movement: experimental artefacts. Clin. Biomech. 11 (2), 90–100.
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253 Cappozzo, A., Della, C.U., Leardini, A., Chiari, L., 2005. Human movement analysis using stereophotogrammetry. Part 1: theoretical background.. Gait. Posture. 21 (2), 186–196. Charlton, I.W., Johnson, G.R., 2006. A model for the prediction of the forces at the glenohumeral joint. Proc. Inst. Mech. Eng. 220 (8), 801–812. Cutti, A.G., Cappello, A., Davalli, A., 2005a. A new technique for compensating the soft tissue artefact at the upper-arm: in vitro validation. J. Mech. Med. Biol. 5, 1– 15. Cutti, A.G., Paolini, G., Troncossi, M., Cappello, A., Davalli, A., 2005b. Soft tissue artefact assessment in humeral axial rotation. Gait. Posture 21 (3), 341–349. Cutti, A.G., Cappello, A., Davalli, A., 2006a. In vivo validation of a new technique that compensates for soft tissue artefact in the upper-arm: preliminary results. Clin. Biomech. 21 (Suppl 1), S13–S19. Cutti, A.G., Garofalo, P., Davalli, A., 2006b. How accurate is the estimation of elbow kinematics using ISB recommended joint coordinate systems? Gait Posture 24 (S2), S36–S37. Cutti, A.G., Giovanardi, A., Rocchi, L., Davalli, A., Sacchetti, R., 2008. Ambulatory measurement of shoulder and elbow kinematics through inertial and magnetic sensors. Med. Biol. Eng. Comput. 46 (2), 169–178. de Groot, J.H., Brand, R., 2001. A three-dimensional regression model of the shoulder rhythm. Clin. Biomech. 16 (9), 735–743. Garofalo, P., Cutti, A.G., Filippi, M.V, Cavazza, S., Ferrari, A., Cappello, A., Davalli, A., 2009. Inter-operator reliability and prediction bands of a novel protocol to measure the coordinated movements of shoulder-girdle and humerus in clinical settings. Med. Biol. Eng. Comput., in press. Johnson, G.R., Stuart, P.R., Mitchell, S., 1993. A method for the measurement of three dimensional scapular movement. Clin. Biomech. 8 (5), 269–273. Karduna, A.R., McClure, P.W., Michener, L.A., Sennett, B., 2001. Dynamic measurements of three-dimensional scapular kinematics: a validation study. J. Biomech. Eng. 123 (2), 184–190. Kontaxis, A., Banerjee, S., Bull, M.J.A., Johnson, G., 2007. Kinematics performance on activities of daily living after reverse shoulder joint replacement, in XXI ISB congress. J. Biomech., S108. Kontaxis, A., Johnson, G.R., 2008. Adaptation of scapula lateral rotation after reverse anatomy shoulder replacement. Comput. Meth. Biomech. Biomed. Eng. 11 (1), 73–80. Krumlinde-Sundholm, L., Holmefur, M., Kottorp, A., Eliasson, A.C., 2007. The Assisting Hand Assessment: current evidence of validity, reliability, and responsiveness to change. Dev. Med. Child Neurol. 49 (4), 259–264. Ludewig, P.M., Behrens, S.A., Meyer, S.M., Spoden, S.M., Wilson, L.A., 2004. Threedimensional clavicular motion during arm elevation: reliability and descriptive data. J. Orthop. Sports Phys. Ther. 34 (3), 140–149. Marchese, S.S., Johnson, G.R., 2000. Measuring the kinematics of the clavicle”. In: 6th International Symposium on the 3D Analysis of Human Movement, Cape Twon, S. Africa. pp. 37–40. Mell, A.G., LaScalza, S., Guffey, P., Ray, J., Maciejewski, M., Carpenter, J.E., Hughes, R.E., 2005. Effect of rotator cuff pathology on shoulder rhythm. J. Should. Elbow Surg. 14 (1 Suppl. S), 58S–64S. Meskers, C.G., van de Sande, M.A., de Groot, J.H., 2007. Comparison between tripod and skin-fixed recording of scapular motion. J. Biomech. 40 (4), 941–946. Murray, I.A., Johnson, G.R., 2004. A study of the external forces and moments at the shoulder and elbow while performing every day tasks. Clin. Biomech. 19 (6), 586–594. Peterson, B., Palmerud, G., 1996. Measurement of upper extremity orientation by video stereometry system. Med. Biol. Eng. Comput. 34 (2), 149–154. Piazza, S.J., Cavanagh, P.R., 2000. Measurement of the screw-home motion of the knee is sensitive to errors in axis alignment. J. Biomech. 33 (8), 1029–1034. Pronk, G.M., van der Helm, F.C., 1991. The palpator: an instrument for measuring the positions of bones in three dimensions. J. Med. Eng. Technol. 15 (1), 15–20. Roux, E., Bouilland, S., Godillon-Maquinghen, A.P., Bouttens, D., 2002. Evaluation of the global optimisation method within the upper limb kinematics analysis. J. Biomech. 35 (9), 1279–1283. Schmidt, R., sselhorst-Klug, C., Silny, J., Rau, G., 1999. A marker-based measurement procedure for unconstrained wrist and elbow motions. J. Biomech. 32 (6), 615– 621. Stokdijk, M., Biegstraaten, M., Ormel, W., de Boer, Y.A., Veeger, H.E., Rozing, P.M., 2000. Determining the optimal flexion-extension axis of the elbow in vivo - a study of interobserver and intraobserver reliability. J. Biomech. 33 (9), 1139– 1145. Teece, R.M., Lunden, J.B., Lloyd, A.S., Kaiser, A.P., Cieminski, C.J., Ludewig, P.M., 2008. Three-dimensional acromioclavicular joint motions during elevation of the arm. J. Orthop. Sports Phys. Ther. 38 (4), 181–190.
253
van Andel, C.J., Wolterbeek, N., Doorenbosch, C.A., Veeger, D.H., Harlaar, J., 2008. Complete 3D kinematics of upper extremity functional tasks. Gait. Posture 27 (1), 120–127. van der Helm, F.C., 1994a. A finite element musculoskeletal model of the shoulder mechanism. J. Biomech. 27 (5), 551–569. van der Helm, F.C., 1994b. Analysis of the kinematic and dynamic behavior of the shoulder mechanism. J. Biomech. 27 (5), 527–550. Veeger, H.E., van der Helm, F.C.T., Rozendal, R.H., 1993. Orientation of the scapula in a simulated wheelchair push. Clin. Biomech. 8 (2), 59–111. Veeger, H.E., Yu, B., An, K.N., Rozendal, R.H., 1997. Parameters for modeling the upper extremity. J. Biomech. 30 (6), 647–652. Veeger, H.E., Magermans, D.J., Nagels, J., Chadwick, E.K., van der Helm, F.C., 2006. A kinematical analysis of the shoulder after arthroplasty during a hair combing task. Clin. Biomech. 21 (Suppl 1), S39–S44. Wu, G., van der Helm, F.C., Veeger, H.E., Makhsous, M., Van, R.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. J. Biomech. 38 (5), 981–992.
Glossary (1) Technical frame (TF): A TF is a coordinate system associated with a body segment. It has normally no repeatable reference to the morphology of the segment and as such has an arbitrary position and orientation with respect to the bone. The placement of markers or sensors that define a TF is usually such as to comply with technical requirement (e.g. minimise soft tissue artefact, enhance visibility and comfort, etc.). For optoelectronic systems a TF is usually constructed by using the instantaneous position of at least three non-aligned superficial markers associated with the bony segment, based on an arbitrary geometric rule. For other measurement systems, e.g. electromagnetic or inertial and magnetic, it is the local coordinate system of the sensor associated with the bony segment. (2) Anatomical frame (AF): An AF is a coordinate system associated with a bony segment. The planes of an AF approximate the frontal, transverse and sagittal anatomical planes of its segment. Its axes (named anatomical axes), are only first-order approximations of the real axes of rotation of the joint(s) that the segment forms. As such, when anatomical axes are assumed to be the axes about which a joint rotates, the resultant joint kinematics can be substantially affected by kinematic cross-talk (Piazza and Cavanagh, 2000).For the upper limb there are recommendations (Wu et al., 2005) for the construction of AF using the position of three non-aligned anatomical landmarks associated with the bony segment. For the construction of the AF the anatomical landmarks are usually calibrated with respect to technical frames. Alternatively an AF can be constructed by a postural pose where an external frame coincides with the anatomical axes as they are described by the ISB. (3) Functional frame (FF): A FF is a coordinate system associated with a segment and is specifically intended to describe the kinematics of a joint formed by the segment. The FF is based on at least one functional axis of rotation of the joint, expressed in a technical (or anatomical) frame associated with the segment.A functional axis of a joint is the axis of rotation of the distal and proximal segments that are forming the joint, when these are actively or passively rotated relative to each other. For a pure hinge joint, the functional axis coincides with the axis of the hinge. For an ‘‘almost hinge” joint (e.g. the elbow and the knee), the functional axis is assumed as the mean axis of rotation, computed from the joint instantaneous axes of rotation. For a ball and socket joint there are no preferential axes of rotation. However, if the distal segment is rotated in a constant plane, the functional axis is defined as the axis perpendicular to the plane of rotation. Functional axes can be computed through a number of algorithms. One of the most commonly used is that based on the estimation of Instantaneous Helical Axes which allows expression of the mean axis of rotation of the joint of interest in the technical (or anatomical) frame of the corresponding segment of interest.With the functional axis expressed in the technical (or anatomical) frame of the segment of interest, FF can then be constructed by (1) using a combination of more functional axes of rotation, or (2) use a combination of the functional and anatomical axes.When computing the kinematics of the joint, the functional axis is assumed to be the axis around which joint rotations occur. This minimises the kinematic cross-talk.
Contents lists available at ScienceDirect
Clinical Biomechanics journal homepage: www.elsevier.com/locate/clinbiomech
Review
A framework for the definition of standardized protocols for measuring upper-extremity kinematics A. Kontaxis a, A.G. Cutti b, G.R. Johnson a, H.E.J. Veeger c,d,* a
Newcastle University, Newcastle upon Tyne, United Kingdom INAIL Prostheses Centre, Vigorso di Budrio, Italy c VU University Amsterdam, Amsterdam, The Netherlands d Delft University of Technology, Delft, The Netherlands b
a r t i c l e
i n f o
Article history: Received 23 July 2008 Accepted 24 December 2008
Keywords: Shoulder Elbow Kinematics Framework Standard
a b s t r a c t Background: Increasing interest in upper extremity biomechanics has led to closer investigations of both segment movements and detailed joint motion. Unfortunately, conceptual and practical differences in the motion analysis protocols used up to date reduce compatibility for post data and cross validation analysis and so weaken the body of knowledge. This difficulty highlights a need for standardised protocols, each addressing a set of questions of comparable content. The aim of this work is therefore to open a discussion and propose a flexible framework to support: (1) the definition of standardised protocols, (2) a standardised description of these protocols, and (3) the formulation of general recommendations. Methods: Proposal of a framework for the definition of standardized protocols. Findings: The framework is composed by two nested flowcharts. The first defines what a motion analysis protocol is by pointing out its role in a motion analysis study. The second flowchart describes the steps to build a protocol, which requires decisions on the joints or segments to be investigated and the description of their mechanical equivalent model, the definition of the anatomical or functional coordinate frames, the choice of marker or sensor configuration and the validity of their use, the definition of the activities to be measured and the refinements that can be applied to the final measurements. Finally, general recommendations are proposed for each of the steps based on the current literature, and open issues are highlighted for future investigation and standardisation. Interpretation: Standardisation of motion analysis protocols is urgent. The proposed framework can guide this process through the rationalisation of the approach. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction In the past 20 years, quantitative motion analysis techniques have provided answers to a wide variety of problems concerning upper-extremity biomechanics. Moreover, different authors have addressed similar problems from different perspectives (Anglin and Wyss, 2000b). At this stage of development, one would expect, therefore, to be able to compare or integrate the results from different studies to draw stronger or broader conclusions. Unfortunately, this is not always possible, due to conceptual and practical differences in the motion analysis protocols used resulting in the application of different techniques or terminologies to define similar measurable quantities (Anglin and Wyss, 2000b). For instance, different authors used different kinematic models of the upper-limb (van der Helm, 1994b; Murray and Johnson, * Corresponding author. Address: VU University Amsterdam, Amsterdam, The Netherlands. E-mail address: [email protected] (H.E.J. Veeger). 0268-0033/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.clinbiomech.2008.12.009
2004), coordinate systems (Anglin and Wyss, 2000a; Cutti et al., 2008), motor tasks (Murray and Johnson, 2004, van Andel et al., 2008), scapula or clavicle tracking or optoelectronic marker setups (Johnson et al., 1993; Karduna et al, 2001; Ludewig et al., 2004; Anglin and Wyss, 2000a). This difficulty highlights a need for standardised protocols for motion analysis of the upper extremity. Although it is almost impossible to define a single, universal protocol able to address all possible questions, at least a certain degree of standardisation should be sought. In particular, it should be possible to agree on general recommendations and to define standardised protocols, each addressing a set of questions of comparable content. The aim of this work is therefore to open a discussion and propose a flexible framework to support: (1) the definition of standardised protocols, (2) a standardised description of these protocols, and (3) the formulation of general recommendations. Through the framework, the authors aim to clear confusion regarding kinematics results obtained from anatomical and functional frames, as well as giving recommendations on how to use these
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
different coordinate systems. Finally, the study aims to expose the lack of knowledge in some areas of upper extremity kinematics where there is poor or no standardisation and so trigger further investigations that will bridge the gap. 2. Methods The proposed framework consists of two nested flowcharts. The former (Section 2.1, Fig. 1) defines what is meant by a motion analysis protocol by pointing out its role in a measurement study. The latter (Section 2.2, Fig. 2a) specifies the steps required to build such a motion analysis protocol, appropriate to the study design. Basic recommendations and current gaps for issues discussed in the literature are also discussed in Section 3 and presented in Fig. 2b. 2.1. Definition of a motion analysis protocol As in any scientific investigation, the following basic steps should be taken to successfully conclude a motion analysis study (Fig. 1). First, there is a need to define the (clinical, ergonomic or sports related) research question (Step 1), which in turn defines whether the study should: (i) involve intra- or inter-subject comparisons; (ii) address specific joints, user function assessment, or biomechanical model; (iii) be longitudinal or cross-sectional. A set of hypotheses to be proven or rejected should be formulated in order to address the research question (Step 2), followed by the definition of the parameters required to test the hypotheses (Step 3). Then the Motion Analysis Protocol should be built in order to measure the parameters over the population of the study (Step 4). The additional steps for this purpose are described in Section 2.2 and illustrated in Fig. 2a. After application of this protocol to the population of the study (Step 5) and a statistical analysis over the parameters measured (Step 6) the hypotheses can be verified/falsified (Step 7) to conclude the study. A motion analysis protocol is therefore the means of measurement of the parameters required to test the hypotheses at the base of the study research question. 2.2. Building a motion analysis protocol When constructing a motion analysis protocol the following six issues (Sections 2.2.1–2.2.6, Fig. 2a) must be addressed: 2.2.1. Joint and/or segment kinematics to be studied It is first necessary to specify whether joint and/or segment kinematics is of interest, and to identify the joints and segments involved. Segment kinematics is defined as the attitude of one bony segment with respect to either a global coordinate system or a nonadjacent bone. Examples of segment kinematics are the attitude of the thorax in the global coordinate of the measurement system or the attitude of the humerus relative to the thorax.
247
Joint kinematics is defined as the relative attitude of two adjacent bony segments. Joint kinematics is for example the attitude of the clavicle relative to the thorax, scapula relative to the clavicle, scapula relative to the thorax, or humerus relative to the scapula. As addressed further in Section 2.2.3, the rotational component of joint kinematics can be described through the decomposition of a rotation matrix as a series of rotations around either predefined anatomical axes or joint functional axes (see the Glossary). The choice of anatomical versus functional axes has important implications in the definition of the protocol and in particular in the definition of anatomical/functional coordinate systems. It is important to note that not all joints/segments of interest may be measurable, depending on the instrumentation or time available to perform the measures. For example: (1) recording of the 3D attitude of the clavicle with non invasive methods requires specialised equipment (Marchese and Johnson 2000); (2) scapula tracking for humerus elevation higher than 120°. requires specialised equipment and/or sufficient time (Johnson et al., 1993; Pronk and van der Helm, 1991). The final list of joints or segments of interest should be formulated after considering all measurement and time constraints. 2.2.2. Mechanical model of joints/Degrees of Freedom for segment kinematics If the focus of the study is joint kinematics, a mechanical (kinematic) equivalent of the joints of interest must be formulated, specifying the number of linkages and the Degrees of Freedom (DoFs) (Fig. 3). The mechanical model must take into account known preferential axes of rotation of the joint. On the other hand, if segment kinematics is the object of the study, it should be specified whether segments are considered to have six DoFs (three rotations plus three translations) or otherwise. 2.2.3. Definition of joint or segment coordinate systems and angles Irrespective of whether joint or segment kinematics are to be studied, local coordinate systems are required to define the attitude of one segment relative to another. For the joints of interest, coordinate systems must be defined consistently with the appropriate mechanical models. The definition of a joint coordinate system typically requires the definition of two coordinate systems, one for the proximal and one for the distal segment forming the joint. To ensure optimal comparability between different studies, it is proposed that these two coordinate systems should be anatomical frames (see Glossary and Fig. 4). However, the relative orientation of the anatomical frames of the proximal and distal segment can be only indicative of the actual joint rotations because the anatomical axes forming these frames are generally only a rough approximation of the real joint axes of rotations. The consequence of the approximation is a joint kinematics affected by kinematic cross-talk (Piazza and Cavanagh, 2000). If the calculated rotations have to approximate joint motion closely, or the measuring system precludes the measurement of bony landmarks, functional frames should be used for the
Fig. 1. Flowchart that describes the basic steps to conclude a motion analysis study, thus pointing about the role of a motion analysis protocol.
248
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
Fig. 2. (a) Flowchart that specifies the steps required to build a motion analysis protocol (left panel); (b) basic recommendations and current gaps for each of the steps (right panel). Analytical references are presented in the corresponding sections indicated on the panel.
segments, in which the axes are defined on the basis of functional axes of rotation (see Glossary and Fig. 4). An example of how the choice of anatomical versus functional frames can substantially affect the measurement of elbow kinematics is shown in Appendix (adapted from Cutti et al. (2008)). If two joints are of interest involving a single segment then two coordinate systems may need to be defined for the same segment, one associated with its proximal and one with its distal end. The use of one single segment coordinate system may preclude the possibility to describe appropriately the kinematics of both joints at the two ends of the segment. This, for example, is the case for the humerus and the description of elbow and glenohumeral joints (Cutti et al., 2008) (Fig. 5).
For segment kinematics, segment coordinate systems should be representative of the anatomy of the bone. As for joint kinematics, in order to ensure maximum comparability of results between studies, these should be expressed using anatomical frames (Wu et al., 2005) For interpretation of either joint or segment kinematics, appropriate sequences of Euler angles must be defined to express the relative orientation of coordinate systems in a clinically meaningful manner. 2.2.4. Marker/sensor set-up Once the anatomical/functional coordinate systems are defined, they must be linked to the markers/sensors of the measurement
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
249
Fig. 3. Example of a simple mechanical (kinematic) equivalent for the upper-limb. In (a) the linkages (black lines) and joints (squares) of interest are pointed out, while in (b) the degrees of freedom of each joint are further specified. Acronyms: Elevation–Depression (El–De), Protraction–retraction (Pr–Re), Internal–External rotation (In–Ex), elbow Flexion–Extension (eFl–Ex), Pronation–Supination (Pn–Su), Radio–Ulnar deviation (Rd–Ul), wrist Flexion–Extension (wFl–Ex).
system. This requires (1) to define marker/sensor locations over the subject’s body, and (2) to establish the relationship between the joint/segment coordinate systems and the technical frames (see Glossary) associated with the markers/sensors. 2.2.5. Activities to be measured The protocol should specify the set of activities to be executed in the study and should differentiate between those used for calibration (e.g., set-up of joint coordinate systems, definition of landmarks, joint centres and rotational axes) and the activities aiming to investigate the function. In addition, there must be (1) a full description of the activities to be executed together with (2) a clear description of how they will be presented to the subjects, and (3) a statement of the number of repetitions. 2.2.6. Kinematics estimation and refinement through mechanical/ signal modelling Mechanical models can be applied to estimate or refine the attitude of specific segments, starting from the attitude of other segments or by applying specific constraints over the measured kinematics (Biryukova et al., 2000). Similarly, authors have proposed specific algorithms to compensate for errors affecting segment attitude (Cutti et al., 2005a; Roux et al., 2002; Schmidt et al., 1999). Only methodologies of proven validity should be applied. If refinements are applied the protocol should include a clear description of the methodology applied. 2.3. Current standardisation and issues to be addressed In the following section the authors have tried to summarise, based on the available literature, some basic recommendations and open issues for the six steps required to define a protocol (Section 2.2). A schematic summary of these recommendations and open issues is presented in Fig. 2b. 2.3.1. Joints/segments of interest Depending on the study and the research question the appropriate joints/segments to be measured should be selected. Note that
Fig. 4. Example of Anatomical Frames (AF) for thorax, humerus and forearm, and of Functional Frames (FF) for humerus and forearm to optimally describe the elbow joint kinematics. The AFs are based on the ISB recommendations (Wu et al., 2005) while the FFs are based on the elbow Flexion–Extension (Fl–Ex) and PronoSupination (Pn–Su) axes of rotations (Cutti et al., 2008). Black dashed lines: segment axes; black dashed-dotted lines: elbow joint axes.
250
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
not all joint/segment kinematics may be measurable (e.g. instrumentation or time restrictions), or need to be measured. General recommendations are therefore difficult in this case. However, in functional studies, even if previous publications focused only on shoulder/elbow (Murray and Johnson, 2004; Veeger et al., 2006), hand position and orientation should also be considered since it frequently determines the accomplishment of a functional task (van Andel et al., 2008). This recommendation is supported by the extensive attention given to the hand in clinical scales assessing function and capacity in children (Krumlinde-Sundholm et al., 2007). In some clinical settings, it can be useful to analyse arm motion only, without entering into the details of the scapula or clavicle (Cutti et al., 2005b; Garofalo et al., 2009). 2.3.2. Mechanical model of joints Model definitions vary from linked segment models to complex musculoskeletal models with different DoFs for each joint. For example Murray and Johnson (2004) defined a nine linkage mechanism to describe only functional shoulder, elbow and hand orientations whereas others used more links to measure and describe also the scapula and clavicle (Charlton and Johnson, 2006; van der Helm, 1994a). As a general recommendation, protocols should refer to published models whenever possible. Otherwise, a detailed review of all the DoFs and of the closed or open loop mechanisms should be given for future reference and comparisons between studies. 2.3.3. Definition of joint/segment coordinate systems and angles In combination with the mechanical model definition, consistent coordinate systems should be defined for the joints/segments of interest. It is proposed that the standardisation proposal of the International Society of Biomechanics-International Shoulder Group (ISB-ISG) (Wu et al., 2005) on anatomical frames and rota-
Fig. 6. Appendix figure: description of (i) a technical frame (based on markers M1, M2 and M3), (ii) an anatomical frame for the proximal humerus (based on the position of anatomical landmarks GH, EL, EM) and (iii) a functional frame for the distal humerus (based on the functional Flexion/Extension axis of elbow and on the humerus longitudinal anatomical axis) (a) frontal view; (b) sugittal view.
tion/decomposition orders for the upper extremity should be followed. For the description of glenohumeral rotations, this order implies gimbal lock at 0° and 180° elevation. If motion around these positions is important, the sequence Flexion–Extension, AbAdduction and Internal–External rotation is recommended for movements in the sagittal plane. For frontal movements, the sequence Ab-Adduction, Flexion–Extension and Internal–External rotation is recommended instead (Cutti et al., 2008). For the use of functional frames, no standard yet exists. It is strongly recommended that this standard, when proposed, should include the link to anatomical frames. This could be based upon information regarding the orientation of joint motion axes relative to anatomical coordinate systems (Stokdijk et al., 2000; Veeger et al., 1997) or by direct measurement of both anatomical landmarks and functional axes. It is important to keep in mind that the use of anatomical frames to describe joint rotations will produce apparent joint motion errors, i.e. kinematic cross-talk. If joint motion needs to be estimated using anatomical frames, then corrections for this difference in alignment can be applied (Cutti et al., 2005a). 2.3.4. Marker/sensor set-up The defined coordinate systems must be linked to the sensors traceable by the measurement systems. Anglin and Wyss (2000b) have presented an extended review on marker set-ups and landmark location. However, to the knowledge of the authors, no methodological study is yet available to recommend the optimum set-up for a specific type of pathology or group of subjects, where movement constraints, loose skin and fat issues can create inaccuracies. It should be noted that marker placement over anatomical landmarks can create large skin artefacts (Cappozzo et al., 1996). Thus it is proposed: Not to attach marker/sensor on anatomical landmarks, instead use technical frames (see Glossary), and refer the anatomical/functional frames to the technical frames.
Fig. 5. Definition of two coordinate systems for the humerus (proximal and distal ends) to optimally describe the glenohumeral and elbow rotations. The relative orientation of the humerus proximal and thorax coordinate systems describes the glenohumeral joint rotations; the relative orientation of the forearm and humerus distal coordinate systems describes the elbow joint rotations. The thorax and proximal humerus coordinate systems follow the ISB recommendations (Wu et al., 2005)while the distal humerus and forearm coordinate systems follow (Cutti et al., 2008).
The latter step can be achieved by either: (i) anatomical calibrations to define anatomical landmarks and thus anatomical axes and frames (Cappozzo et al., 2005) (ii) and/or functional movements to define functional axes and thus functional frames (Biryukova et al., 2000; Cutti et al., 2008) – see
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
Glossary, or (iii) the use of a reference pose position with predefined segment orientation in which all joint angles are predefined (Cutti et al., 2008). If reference positions for the arm are used, it is proposed that a position, in which the elbow and forearm are in their mid position, or 90° Flexion and 90° Pronation, should be used. This position is relatively easily adopted, even within patient groups and in the movement mid range, thus ensuring that possible estimation errors are evenly spread. Scapular movement is also an essential component of arm elevation and is frequently compromised in pathologies (Kontaxis and Johnson, 2008, Mell et al., 2005). Tracking of scapula motion has been performed in normal and pathological shoulders, with a static palpation method and this is probably the most validated, but rather slow, method for clinical measurement (Johnson et al., 1993). Recent studies (Karduna et al., 2001; Meskers et al., 2007) have indicated that tracking scapular motion by attaching a sensor on the acromion is valid for humerus elevation up to approximately 120°., but the technique was validated only in young healthy subjects. In pathological shoulders with compromised glenohumeral motion the validity of the method is still uncertain. Therefore, for the time being, a static palpation technique is still recommended. If scapular tracking is not possible (dynamically or statically), the only remaining option for the estimation of scapula orientation is the application of regression equations. These can be constructed on an individual basis, within a particular experiment and application (Veeger et al., 1993), but more generalised regression equations are available (de Groot and Brand, 2001). Optimal results do, however, still require the measurement of the scapula resting position. Few studies that have included the direct measurement of clavicle kinematics, since it is difficult to track axial rotation in-vivo without the use of invasive methods (Ludewig et al., 2004; Marchese and Johnson, 2000; Teece et al., 2008). As an alternative, clavicle kinematics is estimated by minimising either the changes in length of the conoid ligament (Pronk and van der Helm, 1991) or the axial rotations of the stiff AC ligament (Charlton and Johnson, 2006). Any algorithm that estimates either scapula or clavicle kinematics should be fully described because it may influence the final measured results.
2.3.5. Set of activities to be measured Defining the number and set of activities to be measured is challenging and depends strongly on the research question. Regarding the number of activities, joint function studies should be addressed with a limited number of standard activities, e.g. single joint motions in the standard planes. On the other hand, functional studies should consider the type of sport/ergonomic/ orthopaedic pathology examined. A number of user function studies have recommended different sets of activities in order to challenge several aspects of the joints of interest (Anglin and Wyss, 2000a; Murray and Johnson, 2004; Peterson and Palmerud, 1996; van Andel et al., 2008). As a general recommendation, activities included in validated clinical scales should be preferred, (e.g. Constant Shoulder, SF36, Oxford Shoulder, DASH shoulder/Elbow, Mayo Elbow, SHUEE or AHA) in order to enhance the possibility to correlate the results of the motion analysis study with the clinical conclusions. All the standardised protocols should consider the subject’s comfort and disability taking into account mobility or time constraints. The number of repetitions should be optimised depending on the physical ability of the subjects, but should be enough to ensure that the measurements of motion cycles are truly representative. If possible, at least three representative repetitions should be available.
251
For functional activities, instructions should be verbal or visual but not guided by the investigator. Several studies have presented compensation mechanisms and variations in arm kinematics according to pathology (Kontaxis et al., 2007; Veeger et al., 2006) or even adaptation under high performance sport tasks (Borsa et al., 2008). During functional activities, the kinematic pattern can be important to explain pathologies and disabilities. Changes to environmental set-up can affect the measurement. If changes are unavoidable (e.g. everyday objects, furniture) it is recommended to measure a control group for each set-up. Activities that describe calibration or pose set-ups or define joint rotational axes etc., can be guided (physically or with a device) in order to achieve maximum accuracy of the measurements. 2.3.6. Kinematics estimation and refinement through mechanical modelling In shoulder motion analysis, soft tissue artefacts can corrupt the calculation of some joint angles. In particular there are reports that all axial rotations tend to be substantially underestimated (Cutti et al., 2005a; Schmidt et al., 1999) and they should be analysed with great caution. In particular, while intra-subject comparisons may still be possible, inter-subject comparison may not be valid. Compensation techniques have been developed and/or tested on upper-limb kinematics such as the correction of the humeral and forearm axial rotation (Cutti et al., 2005a, 2006a; Roux et al., 2002; Schmidt et al., 1999). However, these techniques are not yet fully or extensively validated for all upper extremity segments. It thus appears to be too early to formulate specific recommendations. Until then, researchers should be aware of the limited validity of some angles, as already pointed out. 2.4. Parameters analysis As an additional part of the framework, the authors believe that the discussion should lead to the formulation of recommendations for the treatment of certain kinematic output/parameters. As stated in its definition, a motion analysis protocol is the means to measure a set of parameters over the population of the study. The parameters to be computed in the most general case may vary from study to study. However, certain types of parameters are recurrent in upper extremity studies. These are typically: angle patterns versus time, e.g. the glenohumeral Flexion– Extension; angle-angle plots, e.g. the scapula medio-lateral rotation vs humerothoracic Flexion; range of motion. At least for these common parameters, the protocol should include a description of: how to present them graphically; and, for angle-time and angleangle patterns, how to normalize and extract confidence bands. Lastly, scapular attitudes have been presented as many different decompositions and relative to thorax, humerus, or global coordinate systems. To ensure maximal compatibility of data, we suggest that scapula attitude should be presented as the scapulo-thoracic attitude, but always in relation to the thoraco-humeral attitude. This will allow for every other transformation. This specification will much improve the communication and comparison of results.
252
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253
3. Discussion and conclusions
Appendix A
In this paper the authors have tried to summarise the essential steps to be addressed in the definition of motion analysis protocols focusing on (clinical) upper extremity kinematics. In addition, some basic recommendations have been formulated and open problems have been identified. As stated in the introduction, the aim of this paper was not to define one single standardised protocol, but to provide guidelines for protocol developments, comparable to initiatives elsewhere (Bossuyt et al., 2003). Detailed protocols with specific recommendations will have to follow, to measure parameters useful to answer to questions of comparable content. These detailed standardised protocols should be easily developed following the proposed framework which was specifically developed to be both flexible (to adapt to different studies) and detailed in the content to clearly point out the steps required to define a protocol. Both in the framework and in the recommendations, we have tried to point out the difference between segment and joint kinematics, as well as the difference between anatomical and functional frames. This effort was motivated by the feeling of confusion regarding the meaning of the angle patterns obtained by decomposing the relative orientation of the anatomical frames of two adjacent segments. As already pointed out, the axes of anatomical frames are only rough approximations of the real joint axes of rotations. As a consequence, the angles may not closely follow the actual rotations of the joint. Only rotations obtained by decomposing the orientation around functional axes can give indications of the real joint rotations. The confusion that sometimes emerges can also be due to the terminology originally proposed in the ISB recommendations for ‘‘Joint Coordinate Systems”. We think that efforts should be put in reviewing the terminology to enhance clarity. This appears even more urgent since the importance and the attention towards the estimation of functional frames will most probably increase in the next future, due to the diffusion of new measurement systems based in inertial sensors: these systems, in fact, are not (yet) able to measure the position of single anatomical landmarks, but only orientation. By pointing out the essential steps for the definition of a protocol, we hope also to have clarified that a protocol definition is not merely related to ‘‘where to position sensors or markers on the body”, but a much more complex task based on the definition of a kinematic model of the study objective. Finally, we would like to identify the questions that, in the future, will require to be further discussed and addressed:
A.1. Building a technical, an anatomical, and a functional frame: application to the upper arm and to the Humerus (Fig. 6)
(1) the clear distinction between joint angles obtained from anatomical and functional frames; (2) the formulation of guidelines for scapula tracking; (3) the comparison of different set-ups for motion analysis; (4) the problem of soft tissue artefact compensation.
A.1.1. Describing a technical frame on the upper arm Given three markers M1, M2 and M3 attached on the skin of the upper arm, the TF can be constructed as follows:
Y TF ¼ M 2 M1 =kM1 M2 k Z TF ¼ Y TF ^ ðM1 M3 Þ=kX TF ^ ðM1 M 3 Þk X TF ¼ Y TF Z THX The above technical frame has an arbitrary alignment and it could have been defined using a different geometric rule. The markers though are placed in a high visibility position in order to maximise traceability. A.1.2. Humeral anatomical frame (proximal) Following the ISB recommendations (Wu et al., 2005) the anatomical frame H1 of the humerus is constructed based on the centre of rotation of the glenohumeral joint (GH) and the medial (EM) and lateral epicondyles (EM):
Y H1 ¼ ðGH EÞ=kðGH EÞk : longitudinal Z H1 ¼ Y H1 ^ ðEM ELÞ=kY H1 ^ ðEM ELÞk : antero-posterior X H1 ¼ ðY H1 ^ Z H1 Þ=k k : medio-lateral E ¼ ðEL þ EMÞ=2 GH ¼ Origin of the frame The axes of H1 can be used to describe both the glenohumeral joint rotations as well as the elbow joint rotations. While reasonable for the description of the GH joint kinematics, the use of Humeral AF (H1) for the elbow leads to a consisted kinematic cross-talk, which materializes in apparent rotations of the forearm, during pure Flexion–Extensions of the elbow (Cutti et al., 2006b) A.1.3. Humeral functional frame (distal) Following Cutti et al. (2008) the functional frame for the humerus, intended to describe the elbow Flexion–Extension, is constructed as follows:
X HD ¼ V FLEX =kV FLEX k : lateral Z HD ¼ X HD ^ Y H1 =kX HD ^ Y H1 k : posterior Y HD ¼ Z HD ^ X HD =kZ HD ^ X HD k : cranial where VFLEX is the mean Flexion–Extension axis of rotation of the elbow computed through the helical axis algorithm (Veeger et al., 1997), while YH1 is the longitudinal anatomical axis of the humerus (see above). The use of this FF compared to the anatomical frame H1 minimises the kinematic cross-talk at the elbow joint. References
Finally, the authors would reiterate that this paper should be seen as the basis for discussion in the research and clinical communities and eventually lead to an agreed set of protocols and standards.
Acknowledgments The authors gratefully acknowledge the support of Michele Raggi, BEng, INAIL Prostheses Centre, in the preparation of the illustrations. The authors would like to acknowledge the significant contribution to this aspect shoulder biomechanics made by the late Ingram Murray who died at the tragically young age of 35 in December 2008.
Anglin, C., Wyss, U.P., 2000a. Arm motion and load analysis of sit-to-stand, standto-sit, cane walking and lifting. Clin. Biomech. 15 (6), 441–448. Anglin, C., Wyss, U.P., 2000b. Review of arm motion analyses. Proc. Inst. Mech. Eng. [H.] 214 (5), 541–555. Biryukova, E.V., Roby-Brami, A., Frolov, A.A., Mokhtari, M., 2000. Kinematics of human arm reconstructed from spatial tracking system recordings. J. Biomech. 33 (8), 985–995. Borsa, P.A., Laudner, K.G., Sauers, E.L., 2008. Mobility and stability adaptations in the shoulder of the overhead athlete: a theoretical and evidence-based perspective. Sports Med. 38 (1), 17–36. Bossuyt, P.M., Reitsma, J.B., Bruns, D.E., Gatsonis, C.A., Glasziou, P.P., Irwig, L.M., Lijmer, J.G., Moher, D., Rennie, D., de Vet, H.C., 2003. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ 326 (7379), 41–44. Cappozzo, A., Catani, F., Leardini, A., Benedetti, M.G., Croce, U.D., 1996. Position and orientation in space of bones during movement: experimental artefacts. Clin. Biomech. 11 (2), 90–100.
A. Kontaxis et al. / Clinical Biomechanics 24 (2009) 246–253 Cappozzo, A., Della, C.U., Leardini, A., Chiari, L., 2005. Human movement analysis using stereophotogrammetry. Part 1: theoretical background.. Gait. Posture. 21 (2), 186–196. Charlton, I.W., Johnson, G.R., 2006. A model for the prediction of the forces at the glenohumeral joint. Proc. Inst. Mech. Eng. 220 (8), 801–812. Cutti, A.G., Cappello, A., Davalli, A., 2005a. A new technique for compensating the soft tissue artefact at the upper-arm: in vitro validation. J. Mech. Med. Biol. 5, 1– 15. Cutti, A.G., Paolini, G., Troncossi, M., Cappello, A., Davalli, A., 2005b. Soft tissue artefact assessment in humeral axial rotation. Gait. Posture 21 (3), 341–349. Cutti, A.G., Cappello, A., Davalli, A., 2006a. In vivo validation of a new technique that compensates for soft tissue artefact in the upper-arm: preliminary results. Clin. Biomech. 21 (Suppl 1), S13–S19. Cutti, A.G., Garofalo, P., Davalli, A., 2006b. How accurate is the estimation of elbow kinematics using ISB recommended joint coordinate systems? Gait Posture 24 (S2), S36–S37. Cutti, A.G., Giovanardi, A., Rocchi, L., Davalli, A., Sacchetti, R., 2008. Ambulatory measurement of shoulder and elbow kinematics through inertial and magnetic sensors. Med. Biol. Eng. Comput. 46 (2), 169–178. de Groot, J.H., Brand, R., 2001. A three-dimensional regression model of the shoulder rhythm. Clin. Biomech. 16 (9), 735–743. Garofalo, P., Cutti, A.G., Filippi, M.V, Cavazza, S., Ferrari, A., Cappello, A., Davalli, A., 2009. Inter-operator reliability and prediction bands of a novel protocol to measure the coordinated movements of shoulder-girdle and humerus in clinical settings. Med. Biol. Eng. Comput., in press. Johnson, G.R., Stuart, P.R., Mitchell, S., 1993. A method for the measurement of three dimensional scapular movement. Clin. Biomech. 8 (5), 269–273. Karduna, A.R., McClure, P.W., Michener, L.A., Sennett, B., 2001. Dynamic measurements of three-dimensional scapular kinematics: a validation study. J. Biomech. Eng. 123 (2), 184–190. Kontaxis, A., Banerjee, S., Bull, M.J.A., Johnson, G., 2007. Kinematics performance on activities of daily living after reverse shoulder joint replacement, in XXI ISB congress. J. Biomech., S108. Kontaxis, A., Johnson, G.R., 2008. Adaptation of scapula lateral rotation after reverse anatomy shoulder replacement. Comput. Meth. Biomech. Biomed. Eng. 11 (1), 73–80. Krumlinde-Sundholm, L., Holmefur, M., Kottorp, A., Eliasson, A.C., 2007. The Assisting Hand Assessment: current evidence of validity, reliability, and responsiveness to change. Dev. Med. Child Neurol. 49 (4), 259–264. Ludewig, P.M., Behrens, S.A., Meyer, S.M., Spoden, S.M., Wilson, L.A., 2004. Threedimensional clavicular motion during arm elevation: reliability and descriptive data. J. Orthop. Sports Phys. Ther. 34 (3), 140–149. Marchese, S.S., Johnson, G.R., 2000. Measuring the kinematics of the clavicle”. In: 6th International Symposium on the 3D Analysis of Human Movement, Cape Twon, S. Africa. pp. 37–40. Mell, A.G., LaScalza, S., Guffey, P., Ray, J., Maciejewski, M., Carpenter, J.E., Hughes, R.E., 2005. Effect of rotator cuff pathology on shoulder rhythm. J. Should. Elbow Surg. 14 (1 Suppl. S), 58S–64S. Meskers, C.G., van de Sande, M.A., de Groot, J.H., 2007. Comparison between tripod and skin-fixed recording of scapular motion. J. Biomech. 40 (4), 941–946. Murray, I.A., Johnson, G.R., 2004. A study of the external forces and moments at the shoulder and elbow while performing every day tasks. Clin. Biomech. 19 (6), 586–594. Peterson, B., Palmerud, G., 1996. Measurement of upper extremity orientation by video stereometry system. Med. Biol. Eng. Comput. 34 (2), 149–154. Piazza, S.J., Cavanagh, P.R., 2000. Measurement of the screw-home motion of the knee is sensitive to errors in axis alignment. J. Biomech. 33 (8), 1029–1034. Pronk, G.M., van der Helm, F.C., 1991. The palpator: an instrument for measuring the positions of bones in three dimensions. J. Med. Eng. Technol. 15 (1), 15–20. Roux, E., Bouilland, S., Godillon-Maquinghen, A.P., Bouttens, D., 2002. Evaluation of the global optimisation method within the upper limb kinematics analysis. J. Biomech. 35 (9), 1279–1283. Schmidt, R., sselhorst-Klug, C., Silny, J., Rau, G., 1999. A marker-based measurement procedure for unconstrained wrist and elbow motions. J. Biomech. 32 (6), 615– 621. Stokdijk, M., Biegstraaten, M., Ormel, W., de Boer, Y.A., Veeger, H.E., Rozing, P.M., 2000. Determining the optimal flexion-extension axis of the elbow in vivo - a study of interobserver and intraobserver reliability. J. Biomech. 33 (9), 1139– 1145. Teece, R.M., Lunden, J.B., Lloyd, A.S., Kaiser, A.P., Cieminski, C.J., Ludewig, P.M., 2008. Three-dimensional acromioclavicular joint motions during elevation of the arm. J. Orthop. Sports Phys. Ther. 38 (4), 181–190.
253
van Andel, C.J., Wolterbeek, N., Doorenbosch, C.A., Veeger, D.H., Harlaar, J., 2008. Complete 3D kinematics of upper extremity functional tasks. Gait. Posture 27 (1), 120–127. van der Helm, F.C., 1994a. A finite element musculoskeletal model of the shoulder mechanism. J. Biomech. 27 (5), 551–569. van der Helm, F.C., 1994b. Analysis of the kinematic and dynamic behavior of the shoulder mechanism. J. Biomech. 27 (5), 527–550. Veeger, H.E., van der Helm, F.C.T., Rozendal, R.H., 1993. Orientation of the scapula in a simulated wheelchair push. Clin. Biomech. 8 (2), 59–111. Veeger, H.E., Yu, B., An, K.N., Rozendal, R.H., 1997. Parameters for modeling the upper extremity. J. Biomech. 30 (6), 647–652. Veeger, H.E., Magermans, D.J., Nagels, J., Chadwick, E.K., van der Helm, F.C., 2006. A kinematical analysis of the shoulder after arthroplasty during a hair combing task. Clin. Biomech. 21 (Suppl 1), S39–S44. Wu, G., van der Helm, F.C., Veeger, H.E., Makhsous, M., Van, R.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. J. Biomech. 38 (5), 981–992.
Glossary (1) Technical frame (TF): A TF is a coordinate system associated with a body segment. It has normally no repeatable reference to the morphology of the segment and as such has an arbitrary position and orientation with respect to the bone. The placement of markers or sensors that define a TF is usually such as to comply with technical requirement (e.g. minimise soft tissue artefact, enhance visibility and comfort, etc.). For optoelectronic systems a TF is usually constructed by using the instantaneous position of at least three non-aligned superficial markers associated with the bony segment, based on an arbitrary geometric rule. For other measurement systems, e.g. electromagnetic or inertial and magnetic, it is the local coordinate system of the sensor associated with the bony segment. (2) Anatomical frame (AF): An AF is a coordinate system associated with a bony segment. The planes of an AF approximate the frontal, transverse and sagittal anatomical planes of its segment. Its axes (named anatomical axes), are only first-order approximations of the real axes of rotation of the joint(s) that the segment forms. As such, when anatomical axes are assumed to be the axes about which a joint rotates, the resultant joint kinematics can be substantially affected by kinematic cross-talk (Piazza and Cavanagh, 2000).For the upper limb there are recommendations (Wu et al., 2005) for the construction of AF using the position of three non-aligned anatomical landmarks associated with the bony segment. For the construction of the AF the anatomical landmarks are usually calibrated with respect to technical frames. Alternatively an AF can be constructed by a postural pose where an external frame coincides with the anatomical axes as they are described by the ISB. (3) Functional frame (FF): A FF is a coordinate system associated with a segment and is specifically intended to describe the kinematics of a joint formed by the segment. The FF is based on at least one functional axis of rotation of the joint, expressed in a technical (or anatomical) frame associated with the segment.A functional axis of a joint is the axis of rotation of the distal and proximal segments that are forming the joint, when these are actively or passively rotated relative to each other. For a pure hinge joint, the functional axis coincides with the axis of the hinge. For an ‘‘almost hinge” joint (e.g. the elbow and the knee), the functional axis is assumed as the mean axis of rotation, computed from the joint instantaneous axes of rotation. For a ball and socket joint there are no preferential axes of rotation. However, if the distal segment is rotated in a constant plane, the functional axis is defined as the axis perpendicular to the plane of rotation. Functional axes can be computed through a number of algorithms. One of the most commonly used is that based on the estimation of Instantaneous Helical Axes which allows expression of the mean axis of rotation of the joint of interest in the technical (or anatomical) frame of the corresponding segment of interest.With the functional axis expressed in the technical (or anatomical) frame of the segment of interest, FF can then be constructed by (1) using a combination of more functional axes of rotation, or (2) use a combination of the functional and anatomical axes.When computing the kinematics of the joint, the functional axis is assumed to be the axis around which joint rotations occur. This minimises the kinematic cross-talk.