Analysis of the musculoskeletal system of the hand and forearm during a cylinder grasping task

International Journal of Industrial Ergonomics 44 (2014) 535e543

Contents lists available at ScienceDirect

International Journal of Industrial Ergonomics journal homepage: www.elsevier.com/locate/ergon

Analysis of the musculoskeletal system of the hand and forearm during a cylinder grasping task Nicolas Vignais 1, Frédéric Marin* UMR CNRS 7338 Biomechanics and Bioengineering, University of Technology of Compiègne, Research Center, Dct Schweitzer Street, 60200 Compiègne, France

a r t i c l e i n f o

a b s t r a c t

Article history: Received 16 July 2013 Received in revised form 24 January 2014 Accepted 24 March 2014 Available online

Musculoskeletal disorders of the hand are mostly due to repeated or awkward manual tasks in the work environment and are considered a public health issue. To prevent their development, it is necessary to understand and investigate the biomechanical behavior of the musculoskeletal system during the movement. In this study a biomechanical analysis of the upper extremity during a cylinder grasping task is conducted by using a parameterized musculoskeletal model of the hand and forearm. The proposed model is composed of 21 segments, 28 musculotendon units, and 20 joints providing 24 degrees of freedom. Boundary conditions of the model are defined by the three-dimensional coordinates of 43 external markers fixed to bony landmarks of the hand and forearm and tracked with an optoelectronic motion capture system. External marker positions from five healthy participants were used to test the model. A task consisting of closing and opening fingers around a cylinder 25 mm in diameter was investigated. Based on experimental kinematic data, an inverse dynamics process was performed to calculate output data of the model (joint angles, musculotendon unit shortening and lengthening patterns). Finally, based on an optimization procedure, joint loads and musculotendon forces were computed in a forward dynamics simulation. Results of this study assessed reproducibility and consistency of the biomechanical behavior of the musculoskeletal hand system. Relevance to industry: This musculoskeletal model may be employed to predict internal biomechanical parameters during manual handling in the manufacturing industry. Subsequently workplace or tool design may benefit from this process by decreasing the risk of developing work-related musculoskeletal disorders. Ó 2014 Elsevier B.V. All rights reserved.

Keywords: Hand Biomechanics Cylinder grasping Musculoskeletal modeling Motion capture

1. Introduction Musculoskeletal disorders of the hand and forearm are considered a public health issue (Burgess-Limerick, 2007; Picavet and Schouten, 2003). Consequently, understanding how the anatomical structures of the upper limb interact when moving is crucial for designing tools dedicated to industrial manual tasks and decreasing the incidence of work-related musculoskeletal disorders (Dickerson et al., 2007; Vignais et al., 2013). However, direct measurements of musculotendon and joint forces are invasive, and

* Corresponding author. Tel.: þ33 3 44 23 44 23. E-mail addresses: [email protected], [email protected] (N. Vignais), [email protected] (F. Marin). 1 Present address: McMaster Occupational Biomechanics Laboratory, McMaster University, Department of Kinesiology, 1280 Main Street West, Hamilton, Ontario L8S4L8, Canada. http://dx.doi.org/10.1016/j.ergon.2014.03.006 0169-8141/Ó 2014 Elsevier B.V. All rights reserved.

therefore impossible to be performed routinely (Chalfoun et al., 2005; Pfaeffle et al., 1999). As an alternative, methods based on mathematical modeling and computer simulations have been developed in order to analyze the musculoskeletal system of the hand and forearm (Brook et al., 1995; Lemay and Crago, 1996; Sancho-Bru et al., 2001; Wu et al., 2009a). Computer-based musculoskeletal modeling can be divided into two approaches: inverse-based and forward-based simulations (Erdemir et al., 2007). Firstly, the inverse approach estimates musculotendon force by using external data (kinematics, forces, etc.) combined with inverse dynamics and static optimization (Tsirakos et al., 1997). Musculotendon forces assessed by researchers are then compared to electromyographic (EMG) activity patterns to validate their results (Erdemir et al., 2007). Due to complexity and redundancy of its anatomy, several sophisticated inverse models of substructures of the hand and forearm have been suggested (Sancho-Bru et al., 2001; Valero-Cuevas et al., 2003;

536

N. Vignais, F. Marin / International Journal of Industrial Ergonomics 44 (2014) 535e543

Vigouroux et al., 2009). For instance, a three-dimensional biomechanical model of the thumb has been proposed (Wu et al., 2009b). In this case, different virtual scenarios have been designed, e.g. musculotendon forces in a thumb were virtually affected by osteoarthritis and analyzed. This model was verified by comparing the predicted musculotendon moment arms with experimental data (Wu et al., 2009a, 2009b). However, few models have taken into account the whole complexity of the musculoskeletal system of the hand and forearm (Johnson et al., 2009). Chalfoun and colleagues have designed an inverse model of the hand and forearm containing 38 musculotendon units and 17 joints with a total of 24 degrees of freedom (DOF). Based on an optimization method minimizing the square sum of the normalized musculotendon forces, these values were predicted for the closing/opening motion of the hand and pinching (Chalfoun et al., 2004, 2005). Nevertheless, these movements came from a numerical simulation not based on experimental data. Main shortcomings of the inverse approach have been identified as the inadequacy of kinematic models to represent the movement, and inaccuracies of experimental data (Erdemir et al., 2007). Secondly, in the forward approach, an initial set of muscle activations are fed into a forward dynamics model of the musculoskeletal system to estimate the produced movement. Then the solution is compared against experimental data and the process is iterated by updating the muscle activations that best reproduce measured data. Complete musculoskeletal models of the hand and forearm have been proposed using the forward approach to provide

realistic simulations (Albrecht et al., 2003; Li and Zhang, 2009). A musculoskeletal model of the hand and forearm defined by 41 musculotendon units and 16 joints providing 23 DOF has been developed, and was able to compute hand and finger positions with a given set of muscle activations specified by the user (Tsang, 2005). Nevertheless, the forward approach involves a high computational cost (due to multiple integrations to obtain optimal joint kinematics) and is therefore difficult to apply directly to industrial environment where a rapid output is often necessary (Erdemir et al., 2007). Thus the aim of this study is to provide an alternative strategy for the musculoskeletal modeling of the hand and forearm. Based on measured hand motion capture data, this modeling tool includes an inverse-to-forward dynamics simulation in order to estimate subject-specific internal parameters such as musculotendon forces or joint loads. As a case study, the whole procedure has been developed and simulated during a cylinder grasping task. 2. Method 2.1. Overview In order to analyze the biomechanical behavior of the system of the hand and forearm, a three-step process for musculoskeletal modeling is proposed (see Fig. 1). For pre-processing, the LifeMOD (Biomechanics Research Group, San Clemente, CA, USA) plugin has been used and for multi-body dynamics solver, the software

Fig. 1. A three-step process simulation to compute internal parameters. This process includes the musculoskeletal model development (1), the motion capture data experiments (2) and the calculation step with an inverse-to-forward dynamics simulation method using PD (proportional/derivative) and PID (proportional/integral/derivative) controllers (3).

N. Vignais, F. Marin / International Journal of Industrial Ergonomics 44 (2014) 535e543

537

Adams. MSC has been employed (MSC Software Inc., Santa Ana, CA, USA) (Al Nazer et al., 2008). e The first step is the implementation of a highly accurate musculoskeletal model of the hand and forearm. The model is composed of 21 segments, 20 joints providing 24 DOF, and 46 musculotendon units. e The second step deals with the definition of the boundary conditions. The motion capture data have been recorded on five subjects during a cylinder grasping task and imported into the modeling software in the form of motion trajectory markers. e The last step is concerned with calculations. An inverse-toforward dynamics process has been run in order to achieve accurate simulations of musculotendon and joint movements of the hand. Based on the kinematic data from step 2, an inverse dynamics simulation is done first to drive the model. This permits the trainable elements in joints and musculotendon units designed in the step 1 to record joint angulations and musculotendon-contraction histories, respectively. Then these recordable elements are replaced with trained active elements in a subsequent forward dynamics simulation to compute biomechanical parameters like musculotendon forces or joint loads.

2.2. Development of the musculoskeletal model The first step entails defining the geometrical and mechanical behavior of the anatomical structures of the musculoskeletal model, namely anatomical segments, joints, and musculotendon units. All anatomical structures of the model are right-sided. 2.2.1. Anatomical segments The hand and forearm anatomical structures have been modeled as kinematical chains of 21 rigid segments: the forearm composed of the ulna and radius bones, the carpus which includes eight carpal bones (Choi et al., 2010), five metacarpals, five proximal phalanges, four middle phalanges for digits 2 to 4, and five distal phalanges (see Fig. 2). The size and inertial properties of these segments were parameterized by height and weight of the subject based on the GeBOD anthropometric database (Cheng et al., 1994).

Fig. 2. Segments and joints of the hand and forearm model. Segments are listed on the right and joints on the left.

Fig. 3. Direction of axes and localization of MCP, PIP and DIP joints for digit 3 (a) and DOF of the model (b).

Each segment has its own coordinate frame where X-axis is dorsal, Y-axis is distal, and Z-axis is to the right in the anatomical position (Wu et al., 2005). 2.2.2. Joints From an anatomical point of view, most human hand joints are diarthrodial joints. They are defined by cartilage contact surfaces between proximal and distal bones, ligaments and an articular capsule with synovial fluid (Kessler et al., 2007). This complexity can be modeled by a biomimetic mechanical joint with viscoelastic mechanical properties (Marin et al., 2010). Joint centers were localized in the distal tip of each segment by approximating this distal tip as a sphere and by defining its center as the center of the joint. According to the anatomical and biomechanical definition of a diarthrodial joint (Rohen et al., 2006), X-axis corresponds to the dorsal adduction/abduction vector, Y-axis to the distal longitudinal vector, and Z-axis to the lateral flexion/extension vector (see Fig. 3a). Rotational and translational degrees of freedom (DOF) of each joint are based on the functional anatomical description of the joint (see Fig. 3b). In the current model, all segments articulate with a total of 20 joints providing 24 DOF. The PRU joint which enables the movement of pronosupination is modeled by a hinge joint having one DOF. For the RC joint, two rotational DOF, corresponding to flexion/extension and adduction/abduction movements, have been selected. Two DOF have been allocated to the CMC joint of the first digit (for adduction/abduction and flexion/extension movements). There was no DOF in CMC joints of the other digits (Tsang et al., 2005). Two DOF represented the MCP joints (adduction/abduction and flexion/extension movements), and one DOF in each IP joint (see Fig. 3b). The anatomical joint laxity due to ligaments and synovial capsule is characterized by a linear torsional spring and torsional damper applied at each articulation of the model. This torsional spring-damper guides joint movement and keeps it within allowed physical angular limits (Pandy, 2001). In this study, specific stiffness and damping components have been allocated to each articulation (see Table 1). All angular ranges of motion limitations were also documented (Tsang, 2005). 2.2.3. Musculotendon units Altogether the hand and forearm musculoskeletal system consists of 46 musculotendon units (see Table 2). Musculotendon units represent the driving force of the musculoskeletal system by generating segmental motion. Musculotendon unit physiology is

538

N. Vignais, F. Marin / International Journal of Industrial Ergonomics 44 (2014) 535e543

Table 1 Effective joint parameters used in the model. Joint

a

PRU RCa CMC1b MCP1b MCP2c MCP3c MCP4c MCP5c PIP1b PIP2d PIP3d PIP4d PIP5d DIP2d DIP3d DIP4d DIP5d a b c d

Stiffness

Damping

Stiffness limit

N mm/deg

N mm s/deg

N mm/deg

1.130Eþ04 1.130Eþ04 2.62 1.75 12.7 12.7 12.7 12.7 0.87 4.19 4.19 4.19 4.19 2.37 2.37 2.37 2.37

1.130Eþ03 1.130Eþ03 0.262 0.175 1.27 1.27 1.27 1.27 0.087 0.419 0.419 0.419 0.419 0.237 0.237 0.237 0.237

3.390Eþ09 3.390Eþ09 2.620Eþ05 1.750Eþ05 1.270Eþ06 1.270Eþ06 1.270Eþ06 1.270Eþ06 8.700Eþ04 4.190Eþ05 4.190Eþ05 4.190Eþ05 4.190Eþ05 2.370Eþ05 2.370Eþ05 2.370Eþ05 2.370Eþ05

Kaleps et al., 1988. Wu et al., 2009a,b. Howe et al., 1985. Jindrich et al., 2004.

defined by elastic and contractile properties. In order to take into account these specifications, musculotendon units were modeled by a rheological equivalent model with spring and damping elements (Al Nazer et al., 2008). Therefore elastic properties of each musculotendon unit, including tendon path, have been specified as follows: 0.175 N mm/deg for stiffness and 1.751  102 N mm s/deg

Table 2 Physiological cross sectional area (pCSA) assigned to musculotendon units of the model. Musculotendon unit

pCSA (corresponding head or digit) mm2

Biceps Brachii (BB) Brachioradialis (Br) Supinator (SU) Pronator Teres (PT) Flexor Carpi Radialis (FCR) Flexor Carpi Ulnaris (FCU) Flexor Digitorum Superficialis (FDS) Flexor Digitorum Profundus (FDP) Flexor Pollicis Longus (FPL) Flexor Pollicis Brevis (FPB) Extensor Carpi Radialis Longus (ECRL) Extensor Carpi Radialis Brevis (ECRB) Extensor Carpi Ulnaris (ECU) Extensor Digitorum Communis (EDC) Extensor Pollicis Longus (EPL) Extensor Pollicis Brevis (EPB) Extensor Indicis (EI) Extensor Digiti Minimi (EDM) Abductor Pollicis Longus (APL) Abductor Pollicis Brevis (APB) Interossei Palmares (IP) Interossei Dorsales (ID) Lumbricales (LU) Adductor Pollicis (AP) Opponens Pollicis (OP) Opponens Digiti Minimi (ODM) Flexor Digiti Minimi Brevis (FDMB) Abductor Digiti Minimi (ADM) LH

Long Head; SH Short Head; D2 Digit 2; Murray et al., 2000. b Lieber et al., 1992. c Schoenmarklin and Marras, 1993. d Jacobson et al., 1992.

a

250(LH) 133 c 266 b 413 b 199 b 342 b 171(D2) b 177(D2) b 208 d 66 b 146 b 273 b 260 b 52(D2) b 98 d 47 b 56 b 64 b 98 d 68 d 75(D2) d 150(D2) d 11(D2) d 194 d 102 d 110 d 54 d 89

210(SH)

b

D3

Digit 3;

253(D3) 223(D3)

161(D4) 172(D4)

40(D5) 220(D5)

102(D3)

86(D4)

40(D5)

65(D4) 134(D3) 8(D3)

61(D5) 95(D4) 8(D4)

91(D5) 6(D5)

D4

Digit 4;

D5

for damping parameter. Musculotendon action lines have been modeled with slide points. These slide points are rigidly attached to a segment and allow musculotendon units to bend around a geometric feature on that segment (Charlton and Johnson, 2001). For each musculotendon unit, insertion points have been defined in accordance with hand anatomy (Rohen et al., 2006). 2.3. Motion capture data The second step concerns the use of motion capture data as inputs of the musculoskeletal model of the hand and forearm. To this aim, kinematic data of the hand and forearm have been recorded during a 25 mm-cylinder grasping task (closing and opening the hand) at a sampling frequency of 100 Hz. Five subjects participated in this experiment (mean height 178.4  2.1 cm; mean weight 74.3  2.3 kg). All subjects matched to the same upper limb mass properties model in the GeBOD anthropometric database which represented 95% adult human male population (Cheng et al., 1994). The optoelectronic system (VICON, Oxford Metrics Ltd, Oxford, UK) composed of eight T160 cameras was used to track the three-dimensional position of 43 semi spherical reflective markers, each 3 mm in diameter, attached to bony landmarks on hand and forearm (see Fig. 4). During the experiment, the subject was asked to sit upright on a stool behind a table. The stool height and position were adjusted to provide an elbow flexion of 90 deg. The forearm was maintained semiprone on the table and the wrist was kept in neutral position. Fingers were in natural full extension (abduction/adduction not specified), and the palm was facing medially (see Fig. 4a). The subject performed two repetitions of a cylinder grasping task which consisted of closing the hand on a static cylinder (touch only) and then opening the hand to return it to the original position with fingers fully extended (see Video 1). Once the positions of reflective markers were captured, each marker and its trajectory were labeled (see Fig. 4b) and then imported into the simulation software in the form of trajectory markers. These trajectory markers were then attached to massless elements fixed to the musculoskeletal model of the hand and forearm by using spring components. As these massless elements represented the equivalents of kinematical data locations on the model, a first process was run to minimize the offset error between the trajectory marker locations and the placement of massless elements on the model. This process also permitted to compute joint kinematics in accordance with the musculoskeletal model (Dao et al., 2012). Thus, during the inverse dynamics simulation, the model does not exceed joint limits. For the spring components, stiffness and damping components have been fixed to 500 N mm/deg and 50 N mm s/deg respectively. After different simulation tests, these values appeared the most adapted to our kinematical data.

Digit 5.

a

Fig. 4. Optoelectronic reflective markers placed on hand and forearm segments (a) and subsequent labeled kinematic data superimposed on a photo of the protocol (b).

N. Vignais, F. Marin / International Journal of Industrial Ergonomics 44 (2014) 535e543

Supplementary video related to this article can be found at http://dx.doi.org/10.1016/j.ergon.2014.03.006.

539

time step (see Fig. 1). The 25 mm-cylinder grasping movement is finally simulated using a time step of 0.01 s (see Video 1, online data).

2.4. Numerical simulations 2.5. Data analysis Finally, the last step of the musculoskeletal modeling of the hand and forearm concerns the calculation of musculotendon forces and joint loads in accordance with the movement of the hand and forearm recorded during the motion capture experiment. The simulation procedure comprises both inverse and forward dynamics (see Fig. 1). 2.4.1. Inverse dynamics simulation The control method used during the inverse dynamics simulation is the computed torque method (Wasfy and Noor, 2003). This means that kinematic data represented by trajectory markers in the modeling software, serve as input for the simulation. Musculotendon shortening/lengthening patterns are deduced and simultaneously used to compute musculotendon forces Fm. More precisely, from the musculotendon contraction trajectory Tl, musculotendon forces can be obtained by using the formula:

Fm ¼ ðA  Fmax Þ þ Fresting with A ¼

Tl Rl

(1)

(2)

where Fresting is the load on a musculotendon unit in a relaxed state (fixed to 0.45 N by default), Fmax is the product of the physiological cross sectional area pCSA (reported in Table 2) and maximum isometric musculotendon unit stress smax. This last parameter has been adapted to the model of the hand and forearm (smax ¼ 1.4 N/ mm2 for biceps brachii, brachioradialis and supinator; smax ¼ 0.45 N/mm2 for other musculotendon units) (Holzbaur et al., 2005). A represents the muscle activation curve (value between 0 and 1) calculated by the ratio of Tl by Rl, Rl corresponds to the whole musculotendon contraction recorded during the inverse dynamics simulation. In the same way, joint angles are recorded from the model during the inverse dynamics simulation. These data are then used as inputs into the forward dynamics simulation to estimate musculotendon forces and torques. 2.4.2. Forward dynamics simulation During the forward dynamics simulation, musculotendon units are the prime actuators of the model. To guarantee the computational convergence of joint kinematics during the forward dynamics simulation, musculotendon forces are adjusted using PID servo controllers (see Fig. 1). The PID controller minimizes the error between the desired musculotendon contraction trajectory recorded during the inverse dynamics simulation and the instantaneous contraction trajectory obtained from the forward dynamics simulation at each simulation time step (Al Nazer et al., 2008). It has to be noticed that the maximum force of a musculotendon unit is limited by the muscle strength conditioning formula, with A ¼ 1. If PID calculations result in a larger value than this, the force is limited and the PID controller will not force the model to exceed this physiological limit. Joint torques are also used to drive the model during the forward dynamics simulation. In the same manner as for the musculotendon force regulation, a PD controller is applied to reduce the error between the desired joint angle recorded during the inverse dynamics simulation and the instantaneous joint angle calculated from the forward dynamics simulation at each

For the assessment of the reproducibility of results provided by the musculoskeletal model, the two repetitions of the task were compared. More precisely, musculotendon forces and joint loads were computed from the model, time-normalized (1000 frames) and filtered (low pass butterworth filter, cutoff frequency ¼ 5, order ¼ 2) for each repetition. Then intra-individual reproducibility has been evaluated using the Pearson productemoment correlation coefficients between the two repetitions (Popovic et al., 2009). Mean values and standard deviations (n ¼ 5) of these coefficients have been averaged for EDC (extensors) musculotendons, and for FDS and FDP (flexors) musculotendons (see Table 3). Inter-individual reproducibility has also been evaluated. To this aim the mean values and standard deviation of musculotendon forces and joint loads obtained during the second repetition have been used. Then the mean values of the standard deviations from the 5 subjects have been determined and displayed in Table 4. Data from the second repetition have been employed for the subsequent analysis of musculotendon forces and joint loads. 3. Results In the first section, intra- and inter-individual reproducibility of the results provided by the musculoskeletal model is assessed. In the second section, musculotendon forces and joint loads obtained during the cylinder grasping task are presented. These data will be verified in the light of other data from the literature in the discussion section (Blana et al., 2008). 3.1. Intra- and inter-individual reproducibility The intra-individual reproducibility has been assessed during the two repetitions of the task. To this aim, the Pearson producte moment correlation coefficients were computed for musculotendon forces and joint loads. Table 3 shows the mean values and standard deviations of these coefficients obtained from the two repetitions for all participants. In order to test the inter-individual reproducibility, the mean values and standard deviations of musculotendon forces and joint loads have been calculated for the five subjects (see Table 4). 3.2. Musculotendon forces During the cylinder grasping movement, the main muscles exerted were EDC, FDP and FDS (see Fig. 5). The other muscles of the hand and forearm were also activated but not significantly. Table 3 Mean values of correlation coefficients of musculotendon forces (extensor and flexor) and joint loads (CMC, MCP, PIP and DIP) between two cylinder grasping tasks. Data have been averaged for all subjects. Correlation coefficient Musculotendon force Joint load

Extensor Flexor CMC MCP PIP DIP

0.86 0.82 0.46 0.74 0.75 0.43

     

0.18 0.16 0.26 0.21 0.2 0.31

540

N. Vignais, F. Marin / International Journal of Industrial Ergonomics 44 (2014) 535e543

Table 4 Mean values of the standard deviations of the second repetition from 5 subjects for musculotendon forces and joint loads. Standard deviation (mean value) Musculotendon force Joint load

Extensor Flexor CMC MCP PIP DIP

0.77 N 1.03 N 20.97 N 36.84 N 48.2 N 27.91 N

3.3. Joint loads Average CMC, MCP, PIP and DIP joint forces have been computed for all participants and for all digits during the closing/opening motion (see Fig. 6). Interphalangeal joint forces of the thumb have been integrated into PIP joint force calculation. The average distribution of joint force in each finger during the closing/opening movement has also been analyzed (see Fig. 7). Joint force distribution was similar between the second (DIP: 17%, PIP: 21.9%; MCP: 27%; CMC: 34.1%), third (DIP: 18.7%, PIP: 25.1%; MCP: 29.9%; CMC: 26.3%), fourth (DIP: 17.7%, PIP: 25.5%; MCP: 26.9%; CMC: 29.9%) and fifth finger (DIP: 19.6%, PIP: 23.9%; MCP: 22.2%; CMC: 34.2%). The same distal-to-proximal distribution can be observed for the first digit (PIP: 17.6%; MCP: 30.9%; CMC: 51.5%). 4. Discussion Based on measured kinematic data, the current musculoskeletal model of the hand and forearm permitted computation of musculotendon and joint forces during a soft touch task, i.e. a closing/ opening fingers motion around a cylinder. Analysis of the intra and inter-individual reproducibility of the model showed that some variability exists between repetitions of the task and between participants during the closing/opening task. This is consistent with the fact that variability is an intrinsic parameter in human movement analyses (Stergiou and Decker, 2011). Indeed variability in

movement systems helps individuals adapt to the constraints impinging on them across different timescales (Davids et al., 2003). As the only instruction given to participants was to touch the cylinder during the manual task, differences in external force exerted by a subject between movement trials could also explain this variability as musculotendon forces and joint loads depend on the external force applied (Weightman and Amis, 1982). Despite the fact that subjects were instructed to start and finish the task in a fully extended fingers posture small difference of muscle force between the start and the end of the task has been noticed. Considering that the muscle force is based on musculotendon length (Dao et al., 2012), this is the consequence of the fact that fingers postures are more flexed at the beginning of the task. This posture difference between the start and end position after a cyclic motion is well known as the consequence of the neutral zone behavior of joint which induces an involuntary small posture deviation (Marin et al., 2010). Force patterns of musculotendon units obtained during the closing/opening motion of the fingers around the cylinder showed that flexor muscles developed greater forces than extensor muscles on average. Moreover, flexor muscles suddenly decreased their exertion during the last 20% of the movement which correspond to the opening-finger motion. Flexor musculotendon unit of the second finger exerted more force than the other fingers during the movement. Extensor muscles were also activated during the closing/opening motion of the fingers. The EDC musculotendon unit of the third digit reached a maximal value of 3.7 N at 48% of time of the total task. As a verification process, this result can be related to observations from Chalfoun and colleagues’ model (2005) which computed a 2 N musculotendon force exerted by EDC muscle of the second digit during a simulated closing-hand motion. Sancho-Bru et al. (2001) have also modeled the musculoskeletal system of the second digit. The biomechanical behavior of this digit was examined during flexion/extension movements and a musculotendon force of 3 N for EDC muscle was observed (Sancho-Bru et al., 2001). Another model by Brook et al. (1995) reported a higher value (15 N) for EDC muscle of the second digit but it was obtained during a pinching movement (Brook et al., 1995). Our results also

Fig. 5. Muscle force exerted by extensor (A) and flexor (B) muscles during the cylinder grasping action (EDC1,2,3,4 represent musculotendon units of second, third, fourth and fifth digits respectively; likewise for FDP1,2,3,4 and FDS1,2,3,4).

N. Vignais, F. Marin / International Journal of Industrial Ergonomics 44 (2014) 535e543

541

Fig. 6. Average joint forces computed for CMC, MCP, PIP and DIP joints during the closing/opening motion.

demonstrate that during the first half of the movement, the force exerted by extensor muscles is correlated with the finger-flexion motion. This result has already been highlighted in literature (Sancho-Bru et al., 2001) and is related to the cocontraction level of extensor muscles as observed during a maximal power grip task (Goislard de Monsabert et al., 2012). These findings suggest that extensor muscles behave as a brake during free flexion movements of finger joints (Freivalds et al., 2004; Sghaier et al., 2007; ValeroCuevas et al., 1998). The results provided by the current biomechanical model supports this assumption and can be linked to previously reported activities and injuries of extensor muscles related to power grip tasks (Goislard de Monsabert et al., 2012). Joint load analysis revealed that the magnitude of finger joint forces ranged between 22 N and 80 N during the movement (see Fig. 6) which is consistent with values from the literature (Butz et al., 2012; Chao, 1989; Fok and Chou, 2010). Chao and colleagues computed joint forces of the index finger during a static

Fig. 7. Mean joint force distribution among fingers during the movement.

grasping posture and obtained values between 19.8 and 24.5 N for the DIP joint, between 32.5 and 37.8 N for the PIP joint, and between 30.7 and 34.3 N for the MCP joint (Chao, 1989). Our results showed that MCP and PIP joint forces were quite similar during the movement. Brook et al. also found that joint forces at the MCP and PIP joints of digit 2 were equivalent and larger than at the DIP joint, even though it was during a pinching movement (Brook et al., 1995). A small decrease in joint force followed by a constant increase can be observed between 10 and 30% of the movement time for all joint (see Fig. 6). This joint force pattern can be explained by the fact that the closing/opening motion was performed with fingers extended. So it can be suggested that joint forces are lower when fingers are in a natural posture and then rise with increasing finger flexion. Generally, CMC joints supported maximal joint loads followed by MCP joint, PIP joint and DIP joint during the closing/ opening motion. This report is in accordance with most of the studies on joint forces of the hand during power grip actions (Freivalds et al., 2004). Fig. 7 further illustrates the joint force distribution from the DIP to the CMC joints (see Fig. 7). Except for the third finger, CMC joint forces represented the most important contribution although they are generally not considered in the literature. Concerning the thumb, the CMC joint contributes over 50% of joint force, which is in accordance with data from the literature (Domalain et al., 2011; Goislard de Monsabert et al., 2012). Joint torques provided by the model have also been analyzed and demonstrate maximal torque at the MCP joints during the grasping phase of movement. This report can be linked to the work made by Kargov and colleagues who analyzed the grip force distribution in natural hands during a cylinder grasping task. They reported that the highest joint torques were measured at the MCP joints during grasping (Kargov et al., 2004). In the current study, the behavior of the musculoskeletal system of the hand and forearm was investigated during a cylinder grasping task. In the past, this manual task has been selected to represent industrial applications (Dubrowski and Carnahan, 2004; Harih and Dolsak, 2013; Miyata et al., 2012). Indeed grasping a

542

N. Vignais, F. Marin / International Journal of Industrial Ergonomics 44 (2014) 535e543

cylinder involves typical constraints from an industrial manual gestures: all fingers are moving voluntarily and concurrently, muscle contractions are anisometric and the aim of the task is to grasp a hand tool, i.e. a 25 mm-cylinder. Thus it represents one of the most elemental multi-finger manipulative and gestic movements (Braido and Zhang, 2004). By analyzing this task, our model provided an insight into future explorations for the prevention of musculoskeletal disorders of the hand and for the design of hand tools. 5. Limitations In this study we presented a musculoskeletal model of the hand and forearm which can be individually adapted. Segments properties are parameterized based on an anthropometric database and residual geometric differences between the model and the subject are minimized by using a spring component system. In its current form, the musculoskeletal model of the hand and forearm only considers measured kinematic data to drive the model although some specific force devices could be used to integrate measured kinetic data as inputs to the model (Goislard de Monsabert et al., 2012). Despite the fact that EMG signals of forearm muscles are hardly accessible (Chalfoun et al., 2005), this kind of data might also be inputted in combination with the inverse dynamics approach to estimate muscle forces across a series of joints (Erdemir et al., 2007). In the current study, joint torques values appeared overestimated (maximal joint torques range between 130 N/m and 235 N/m for MCP joints). This can be due to the linear modeling of the mechanical behavior of the ligament-capsule structure. Indeed, in biomechanical literature, it is well established that a non-linear model which takes into account the neutral zone is more appropriate to represent the mechanical behavior of an anatomical structure (Marin et al., 2010). The present model is also limited in the field of personalization and validation of a mechanical behavior (Veeger, 2011). However, personalization generally involves complex and additive measurements based on medical imaging (Dao et al., 2012). Consequently, musculoskeletal modeling based on parameterized geometry provides a first insight into musculotendon forces and joint loads and this method remains an acceptable compromise to obtain internal values of a musculoskeletal system for industrial applications (Rasmussen, 2011). 6. Conclusions In this study, we present a musculoskeletal model of the hand and forearm with 21 segments, 20 joints, providing 24 DOF and 46 musculotendon units. The boundary condition of the model is deduced from the motion capture of 43 reflective markers fixed to the hand during a closing/opening fingers task. Results are consistent with literature and they improve our understanding of musculotendon forces and joint loads of the hand during movement. The present musculoskeletal model can be used as a pertinent tool for investigating, preventing and documenting potential musculoskeletal disorders risk during various industrial manual tasks by quantifying musculotendon forces and joint loads. Other practical applications of this model could concern the design of specific rehabilitation exercises or the ergonomic evaluation of industrial manual devices. The present work should be considered as a preliminary three-dimensional musculoskeletal model of the hand for industrial applications. Future investigations will focus on biomechanical data fusion like physiological signals or external loads in order to refine the subject specificity and anthropometry.

Acknowledgments The authors would like to thank all the volunteers that participated in this study. We also gratefully acknowledge the review contribution of Alex MacIntosh. The research leading to these results has received funding from the European Union FP7 2007e 2013 grant agreement n 248290 e Project COGNITO. References Al Nazer, R., Klodowski, A., Rantalainen, T., Heinonen, A., Sievänen, H., Mikkola, A., 2008. Analysis of dynamic strains in tibia during human locomotion based on flexible multibody approach integrated with magnetic resonance imaging technique. Multibody Syst. Dyn. 20, 287e306. Albrecht, I., Haber, J., Seidel, H.P., 2003. Construction and Animation of Anatomically Based Human Hand Models. Proceedings of the 2003 ACM SIGGRAPH/Eurographics Symposium on Computer Animation, Switzerland. Blana, D., Hincapie, J.G., Chadwick, E.K., Kirsch, R.F., 2008. A musculoskeletal model of the upper extremity for use in the development of neuroprosthetic systems. J. Biomech. 41 (8), 1714e1721. Braido, P., Zhang, X., 2004. Quantitative analysis of finger motion coordination in hand manipulative and gestic acts. Hum. Mov. Sci. 22, 661e678. Brook, N., Mizrahi, J., Shoham, M., Dayan, J., 1995. A biomechanical model of index finger dynamics. Med. Eng. Phys. 17, 54e63. Burgess-Limerick, R., 2007. Ergonomics for manual tasks. In: Mayhew, C. (Ed.), Australian Master of OHS and Environment Guide. CCH Australia, North Ryde, pp. 261e278. Butz, K.D., Merrell, G., Nauman, E.A., 2012. A biomechanical analysis of finger joint forces and stresses developed during common daily activities. Comput. Method. Biomec. 15, 131e140. Chalfoun, J., Renault, M., Younes, R., Ouezdou, F., 2004. Muscle Forces Prediction of the Human Hand and Forearm System in Highly Realistic Simulation. Proceedings of the 2004 IEEE/RSJ International Conference on Intelligent Robots and Systems, Japan. Chalfoun, J., Younes, R., Renault, M., Ouezdou, F., 2005. Forces, activation and displacement prediction during free movement in the hand and forearm. J. Robotic Syst. 22, 653e660. Chao, E.Y., 1989. Biomechanics of the Hand: a Basic Research Study. World Scientific Publishing Company, Singapore. Charlton, I.W., Johnson, G.R., 2001. Application of spherical and cylindrical wrapping algorithms in a musculoskeletal model of the upper limb. J. Biomech. 34, 1209e1216. Cheng, H., Obergefell, L., Rizer, A., 1994. Generator of Body Data (GEBOD) Manual. DTIC Document, Armstrong Lab., Boston, Rep. AL/CF-TR-1994-0051, WrightPatterson Air force Base OH. Choi, J.H., Park, S.-W., Baek, S.-Y., Lee, K., 2010. Evaluation of handheld products by computing user hand fatigue. Simul. Model. Pract. Th. 18, 230e239. Dao, T., Marin, F., Pouletaut, P., Charleux, F., Aufaure, P., Ho Ba Tho, M., 2012. Estimation of accuracy of patient-specific musculoskeletal modelling: case study on a post-polio residual paralysis subject. Comput. Method. Biomec. 15, 745e751. Davids, K., Glazier, P., Araujo, D., Bartlett, R., 2003. Movement systems as dynamical systems: the functional role of variability and its implications for sports medicine. Sports Med. 33, 245e260. Dickerson, C.R., Chaffin, D.B., Hughes, R.E., 2007. A mathematical musculoskeletal shoulder model for proactive ergonomic analysis. Comput. Method. Biomec. 10, 389e400. Domalain, M.F., Seitz, W.H., Evans, P.J., Li, Z.M., 2011. Biomechanical effect of increasing or decreasing degrees of freedom for surgery of trapeziometacarpal joint arthritis: a simulation study. J. Orthop. Res. 29, 1675e1681. Dubrowski, A., Carnahan, H., 2004. Grip force when grasping moving cylinders. Int. J. Ind. Ergon. 34, 69e76. Erdemir, A., McLean, S., Herzog, W., van den Bogert, A.J., 2007. Model-based estimation of muscle forces exerted during movements. Clin. Biomech. 22, 131e154. Fok, K.S., Chou, S.M., 2010. Development of a finger biomechanical model and its considerations. J. Biomech. 43, 701e713. Freivalds, A., 2004. Biomechanics of the Upper Limbs: Mechanics, Modelling and Musculoskeletal Injuries. CRC Press, New York. Goislard de Monsabert, B., Rossi, J., Berton, E., Vigouroux, L., 2012. Quantification of hand and forearm muscle forces during a maximal power grip task. Med. Sci. Sport Exerc. 44, 1906e1916. Harih, G., Dolsak, B., 2013. Tool-handle design based on a digital human hand model. Int. J. Ind. Ergon. 43, 288e295. Holzbaur, K.R., Murray, W.M., Delp, S.L., 2005. A model of the upper extremity for simulating musculoskeletal surgery and analyzing neuromuscular control. Ann. Biomed. Eng. 33, 829e840. Howe, A., Thompson, D., Wright, V., 1985. Reference values for metacarpophalangeal joint stiffness in normals. Ann. Rheum. Dis. 44, 469e476. Jacobson, M.D., Raab, R., Fazeli, B.M., Abrams, R.A., Botte, M.J., Lieber, R.L., 1992. Architectural design of the human intrinsic hand muscles. J. Hand Surg.-Am. 17, 804e809. Jindrich, D.L., Balakrishnan, A.D., Dennerlein, J.T., 2004. Finger joint impedance during tapping on a computer keyswitch. J. Biomech. 37, 1589e1596.

N. Vignais, F. Marin / International Journal of Industrial Ergonomics 44 (2014) 535e543 Johnson, E., Morris, K., Murphey, T., 2009. A variational approach to strand-based modeling of the human hand. Spr. Tra. Adv. Robot. 57, 151e166. Kaleps, I., White, R.P., Beecher, B.M., Whitestone, J., Obergefell, L.A., 1988. Measurement of HYBRIDIII Dummy properties and Analytical simulation data Base Development. Armstrong Aerospace Medical Research Laboratory Report no. AAMRL-TR-88-005. Kargov, A., Pylatiuk, C., Martin, J., Schulz, S., Döderlein, L., 2004. A comparison of the grip force distribution in natural hands and in prosthetic hands. Disabil. Rehabil. 26, 705e711. Kessler, O., Dürselen, L., Banks, S., Mannel, H., Marin, F., 2007. Sagittal curvature of total knee replacements predicts in vivo kinematics. Clin. Biomech. 22, 52e58. Lemay, M.A., Crago, P.E., 1996. A dynamic model for simulating movements of the elbow, forearm, and wrist. J. Biomech. 29, 1319e1330. Li, K., Zhang, X., 2009. A novel two-stage framework for musculoskeletal dynamic modeling: an application to multifingered hand movement. IEEE T. Bio-Med. Eng. 56, 1949e1957. Lieber, R.L., Jacobson, M.D., Fazeli, B.M., Abrams, R.A., Botte, M.J., 1992. Architecture of selected muscles of the arm and forearm: anatomy and implications for tendon transfer. J. Hand Surg. Am. 17, 787e798. Marin, F., Hoang, N., Aufaure, P., Ho Ba Tho, M.-C., 2010. In vivo intersegmental motion of the cervical spine using an inverse kinematics procedure. Clin. Biomech. 25, 389e396. Miyata, N., Shimizu, Y., Motoki, Y., Maeda, Y., Mochimaru, M., 2012. Hand MoCap using an individual model with a skeleton and a surface skin. Int. J. Hum. Factor. Model. Simul. 3, 147e168. Murray, W.M., Buchanan, T.S., Delp, S.L., 2000. The isometric functional capacity of muscles that cross the elbow. J. Biomech. 33, 943e952. Pandy, M.G., 2001. Computer modeling and simulation of human movement. Annu. Rev. Biomed. Eng. 3, 245e273. Pfaeffle, J.H., Fischer, K.J., Manson, T.T., Tomaino, M.M., Herndon, J.H., Woo, S.L., 1999. A new methodology to measure load transfer through the forearm using multiple universal force sensors. J. Biomech. 32, 1331e1335. Picavet, H., Schouten, J., 2003. Musculoskeletal pain in the Netherlands: prevalences, consequences and risk groups, the DMC3-study. Pain 102, 167e178. Popovic, N., Williams, S., Schmitz-Rode, T., Rau, G., Disselhorst-Klug, C., 2009. Robotbased methodology for a kinematic and kinetic analysis of unconstrained, but reproducible upper extremity movement. J. Biomech. 42, 1570e1573. Rasmussen, J., 2011. Challenges in human body mechanics simulation. Proc. IUTAM 2, 176e185. Rohen, J.W., Yokochi, C., Lütjen-Drecoll, E., Romrell, L.J., 2006. Color Atlas of Anatomy: a Photographic Study of the Human Body, sixth ed. Lippincott Williams & Wilkins, Philadelphia. Sancho-Bru, J.L., Perez-Gonzalez, A., Vergara-Monedero, M., Giurintano, D., 2001. A 3-D dynamic model of human finger for studying free movements. J. Biomech. 34, 1491e1500.

543

Schoenmarklin, R.W., Marras, W.S., 1993. Dynamic capabilities of the wrist joint in industrial workers. Int. J. Ind. Ergon. 11, 207e224. Sghaier, A., Romdhane, L., Ouezdou, F., 2007. Biomechanical Analysis of the Normal and Reconstructed Human Hand: Prediction of Muscle Forces in Pinch and Grasp. Proceedings of the 12th IFToMM World Congress, France. Stergiou, N., Decker, L.M., 2011. Human movement variability, nonlinear dynamics, and pathology: is there a connection? Hum. Mov. Sci. 30, 869e888. Tsang, H., 2005. Helping Hand: an Anatomically Accurate Inverse Dynamics Solution for Unconstrained Hand Motion. Ph.D. dissertation. Dept. Comp. Sci., Toronto Univ., Toronto. Tsang, W., Singh, K., Fiume, E., 2005. Helping Hand: an Anatomically Accurate Inverse Dynamics Solution for Unconstrained Hand Motion. Proceedings of the 2005 ACM SIGGRAPH/Eurographics Symposium on Computer Animation, USA. Tsirakos, D., Baltzopoulos, V., Bartlett, R., 1997. Inverse optimization: functional and physiological considerations related to the force-sharing problem. Crit. Rev. Biomed. Eng. 25, 371. Valero-Cuevas, F.J., Johanson, M.E., Towles, J.D., 2003. Towards a realistic biomechanical model of the thumb: the choice of kinematic description may be more critical than the solution method or the variability/uncertainty of musculoskeletal parameters. J. Biomech. 36, 1019e1030. Valero-Cuevas, F.J., Zajac, F.E., Burgar, C.G., 1998. Large index-fingertip forces are produced by subject-independent patterns of muscle excitation. J. Biomech. 31, 693e704. Veeger, D.H., 2011. “What if”: the use of biomechanical models for understanding and treating upper extremity musculoskeletal disorders. Man. Ther. 16, 48e50. Vignais, N., Miezal, M., Bleser, G., Mura, K., Gorecky, D., Marin, F., 2013. Innovative system for real-time ergonomic feedback in industrial manufacturing. Appl. Ergon. 44, 566e574. Vigouroux, L., Domalain, M., Berton, E., 2009. Comparison of tendon tensions estimated from two biomechanical models of the thumb. J. Biomech. 42, 1772e1777. Wasfy, T.M., Noor, A.K., 2003. Computational strategies for flexible multibody systems. Appl. Mech. Rev. 56, 553e614. Weightman, B., Amis, A.A., 1982. Finger joint force predictions related to design of joint replacements. J. Biomed. Eng. 4, 197e205. Wu, G., Van der Helm, F.C., Veeger, H., Makhsous, M., Van Roy, P., Anglin, C., Nagels, J., Karduna, A.R., McQuade, K., Wang, X., 2005. ISB recommendation on definitions of joint coordinate systems of various joints for the reporting of human joint motion e part II: shoulder, elbow, wrist and hand. J. Biomech. 38, 981e992. Wu, J.Z., An, K.N., Cutlip, R.G., Andrew, M.E., Dong, R.G., 2009a. Modeling of the muscle/tendon excursions and moment arms in the thumb using the commercial software anybody. J. Biomech. 42, 383e388. Wu, J.Z., Li, Z.-M., Cutlip, R.G., An, K.-N., 2009b. A simulating analysis of the effects of increased joint stiffness on muscle loading in a thumb. Biomed. Eng. Online 8, 41.